Switchgear safety monitoring method based on holographic data combined with ultrasonic signal diagnosis

Abstract

The present invention relates to a switch cabinet safety monitoring method based on holographic data combined with ultrasonic signal diagnosis, including: collecting holographic data of the switch cabinet; using a completely noise-assisted aggregated empirical mode decomposition method to decompose the ultrasonic signal and the time series data of the temperature of each phase moving contact respectively; calculating the correlation between each physical quantity of the switch cabinet and the intrinsic mode component of each phase moving contact temperature and the autocorrelation coefficient between the intrinsic mode components of the ultrasonic signal; screening out the main relevant physical quantities of each phase moving contact temperature and the fault ultrasonic signal for auxiliary diagnosis; using a BiLSTM model to predict the intrinsic mode components of each phase moving contact temperature; synthesizing the intrinsic mode components of the temperature, and according to the predicted value combined with the auxiliary fault diagnosis of the ultrasonic signal, early warning of the safety risk of the switch cabinet. The present invention realizes real-time detection and alarm of high-voltage switch cabinet groups, reduces the accident rate of closed high-voltage switch cabinets, and improves the reliability of the power system.

Description

Technical Field

[0001] The invention belongs to the field of power transmission and transformation control, and in particular relates to a switch cabinet safety monitoring method based on holographic data combined with ultrasonic signal diagnosis. Background Art

[0002] A large number of 10kV high-voltage switchgears are generally arranged indoors in substations with voltage levels of 220kV and above, forming a 10kV high-voltage switchgear group. The high-voltage switchgear is in a closed environment of high voltage, high current and strong magnetic field for a long time, and the high-voltage switchgear generally adopts a closed structure. The staff cannot directly observe the specific conditions of the moving contacts inside the switchgear, the three-phase inlet of the busbar room and other heat-prone parts with the naked eye. The high-voltage switchgear is prone to internal overheating during long-term working projects. The main reason is that the temperature of the moving contacts of the high-voltage switchgear increases due to long-term mechanical movement and contact ablation, causing oxidation of the contact, and increasing the contact resistance. The temperature rises further, and local welding or sparks or even arc discharge occurs, which reduces the insulation level and eventually causes damage to electrical equipment. As equipment to ensure the stable operation of substations, 10kV high-voltage switchgears are numerous and prone to heat. The long-term safe, stable and reliable operation of 10kV high-voltage switchgear directly affects the safety of substations and the power system in which they are located.

[0003] Therefore, the research is to monitor and alarm the temperature, pressure and other data and partial discharge phenomena in the heat-prone parts of the enclosed high-voltage switchgear. Summary of the invention

[0004] The purpose of the present invention is to provide a switch cabinet safety monitoring method based on holographic data combined with ultrasonic signal diagnosis in view of the above problems, and to use the Internet of Things device to collect physical quantities such as temperature and pressure in the high-voltage switch cabinet and ultrasonic signals to obtain holographic data of the high-voltage switch cabinet; to use the empirical mode decomposition method to decompose the time series data of the switch cabinet moving contact temperature and the ultrasonic signal into multiple intrinsic mode components with different frequency domain characteristics; to calculate the correlation between each physical quantity of the switch cabinet and the intrinsic mode components obtained by the decomposition and the correlation coefficient between the intrinsic mode components of the ultrasonic signal and the original signal, and to screen out the main relevant characteristic quantities affecting the moving contact temperature and the intrinsic mode components of the fault ultrasonic signal; when safety risks such as overheating occur, different physical quantities change at different speeds, that is, different physical quantities have different derivatives over time, and the propagation learning ability of the BiLSTM model in the forward and backward directions is used to capture the law of physical quantity changes, and the changing trend of the moving contact temperature is identified in advance, and the ultrasonic signal auxiliary diagnosis system is combined to discover the safety risks of the switch cabinet operation as early as possible and issue an alarm so that the operation and maintenance personnel can take measures in time to ensure power safety and reduce losses.

[0005] The technical solution of the present invention is a switch cabinet safety monitoring method based on holographic data combined with ultrasonic signal diagnosis, comprising the following steps:

[0006] Step 1: Collect the time series data of temperature and pressure of the switch cabinet and the ultrasonic signal of arc or corona phenomenon of the shell to obtain the holographic data of the switch cabinet;

[0007] Use temperature sensors to collect the temperature of the moving contacts of phases A, B, and C of the switch cabinet respectively; collect the temperature of the static contacts of phases A, B, and C and the temperature of the top and back of the switch cabinet respectively; use pressure sensors to monitor the pressure at the contacts of phases A, B, and C in the contact box; use four ultrasonic sensors to detect ultrasonic signals of arc or corona phenomena in the switch cabinet housing;

[0008] Step 2: The Complete Ensemble Empirical Mode Decomposition with adaptive noise (CEEMDAN) method is used to decompose the time series data of the moving contact temperature of phases A, B, and C and the ultrasonic signal into multiple eigenmode components;

[0009] Step 3: Taking the eigenmode components of the temperature of each phase moving contact of the switch cabinet as the target variable, the Spearman correlation analysis is performed on the holographic data of the switch cabinet to calculate the correlation between each physical quantity of the switch cabinet and the eigenmode components of the temperature of each phase moving contact; the correlation coefficients of the eigenmode components of the ultrasonic signal and the original signal are calculated respectively;

[0010] Step 4: According to the correlation with the moving contact temperature, the main related physical quantities of the moving contact temperature of each phase are screened out; according to the correlation coefficient with the original signal, the intrinsic mode component of the fault ultrasonic signal is screened out;

[0011] Step 5: Establish an ultrasonic signal auxiliary diagnosis system, calculate the wavelet packet energy characteristic spectrum of the fault ultrasonic signal eigenmode component screened out in step 4, and then determine whether a partial discharge fault occurs in the switch cabinet based on the ultrasonic signal characteristics;

[0012] Step 6: Establish a BiLSTM model, use the main relevant physical quantities of the moving contact temperature obtained in step 4 as the input of the BiLSTM model, and use the BiLSTM model to predict the intrinsic mode components of the moving contact temperature of each phase; synthesize the intrinsic mode components of the moving contact temperature obtained by the BiLSTM model to obtain the predicted value of the moving contact temperature;

[0013] Step 7: Issue an early warning of the safety risk of the switch cabinet based on the predicted value of the moving contact temperature of the switch cabinet and ultrasonic-assisted diagnosis.

[0014] Preferably, wireless temperature sensors are installed at the static contacts and moving contacts of the three-phase A, B, and C inlet ports in the high-voltage switch cabinet, and at the top and back of the high-voltage switch cabinet, respectively; wireless pressure sensors are installed at the contacts in the contact box; and ultrasonic sensors are installed on the surface of the switch cabinet. The above-mentioned wireless sensors transmit the detected holographic data of the switch cabinet, such as the temperature and pressure of the switch cabinet, and the ultrasonic signals of the arc or corona phenomenon of the shell to the central server through wireless communication. The database of the central server includes a MySQL relational database and a Redis system for storing the holographic data of the switch cabinet. The MySQL database is used to store and query a large amount of data with typical relational characteristics, such as temperature, pressure, and ultrasonic signals. The Redis system is a non-relational database that implements storage according to keyword-value, and supports the storage and synchronization of data types such as strings, linked lists, and collections.

[0015] In step 2, the time series data of the moving contact temperature of the A, B, and C phase inlets collected within the time interval and the ultrasonic signal are decomposed using the complete noise-assisted aggregation empirical mode decomposition method to obtain the intrinsic mode components with different frequency domain characteristics. The specific process includes:

[0016] S1: Add the positive and negative white noise pairs to the original data sequence x(t), and use the empirical mode decomposition method to decompose and obtain the first modal component IMF1:

[0017]

[0018] The residual component is:

[0019]

[0020] Where N represents the sequence length; r1(t) represents the first residual component;

[0021] S2: Add positive and negative paired Gaussian white noise to the residual component r1(t) to obtain a new sequence, and perform empirical mode decomposition to obtain the second modal component IMF2:

[0022]

[0023] Where E k () represents the kth IMF component obtained through empirical mode decomposition, η j (t) is the jth Gaussian white noise sequence; ε1 represents the noise coefficient;

[0024] The residual component is:

[0025]

[0026] Where r2(t) represents the second residual component;

[0027] S3: Similarly to step S2, decompose to obtain the kth residual component:

[0028]

[0029] Where r k-1 (t), r k (t) denotes the k-1th and kth residual components respectively;

[0030] Repeatedly use empirical mode decomposition to decompose the sequence to be processed and derive the k+1th modal component IMF k+1 :

[0031]

[0032] S4: Repeat step S3 until the residual component obtained is a monotonic function and is not suitable for further decomposition, and then end the decomposition process; assuming that the number of modal components obtained is L, the original sequence can be expressed as:

[0033]

[0034] Where r(t) is the final residual component.

[0035] In step 6, the specific process of constructing and using the BiLSTM model to obtain the predicted value of the moving contact temperature includes:

[0036] 1) The intrinsic mode components IMF1-IMF6 and the residual component r of the time series of the moving contact temperatures of phases A, B, and C obtained in step 2 are respectively used as the prediction variables of the BiLSTM model, and the main relevant physical quantities of the moving contact temperature selected in step 4 are used as the input of the BiLSTM model to construct the BiLSTM model;

[0037] Predictors for the BiLSTM model:

[0038] f(x)=ω T x+b (8)

[0039] Where ω represents the weight matrix, b represents the bias term, and x represents the input;

[0040] The collected holographic data of the switch cabinet is used as the training data set of the BiLSTM model for training the BiLSTM model;

[0041] 2) Set the error calculation interval (ε - ,ε + ), where ε - , ε + are the lower and upper limits of the error value, respectively. When the error value is greater than the error calculation interval, the error loss value is calculated;

[0042] 3) In the first LSTM layer of the BiLSTM model, the error is forward propagated to learn from historical data; in the second LSTM layer, the error is backward propagated to learn from future data; the BiLSTM model is trained in both positive and negative directions using the holographic data of the switchgear, so that the BiLSTM model can fully and effectively learn the load information in the holographic data of the switchgear;

[0043] 4) According to the prediction variables of the trained BiLSTM model, the intrinsic mode components IMF1-IMF6 and the residual component r of the time series of the moving contact temperatures of phases A, B, and C are obtained, and the predicted values ​​of the moving contact temperatures of phases A, B, and C are obtained by synthesis; the prediction accuracy of the BiLSTM model is evaluated by using the mean absolute error MAE, root mean square error RMSE, and mean absolute error percentage MAPE.

[0044] The first and second LSTM layers of the BiLSTM model use the following gating structure to control the transmission of information to the cell units through the information selection method:

[0045] Forget Gate:

[0046] f t =σ(W f ·[h t-1 ,x t ]+b f ) (9)

[0047] Where f t represents the output of the forget gate, W f represents the weight matrix of the forget gate, σ( ) represents the activation function; b f represents the bias term of the forget gate; h t-1 represents the hidden variable at time t-1; x t represents the input at time t;

[0048] The forget gate reads h t-1 and x t , outputs a value of 0 to 1 to determine whether the information is transmitted. An output value of 0 indicates that the information is completely discarded, and an output value of 1 indicates that the information is completely retained and transmitted;

[0049] Input Gate:

[0050] i t =σ(W i ·[h t-1 ,x t ]+b i ) (10)

[0051]

[0052] Where i t represents the output of the input gate, W i represents the weight matrix of the input gate, b i represents the bias term of the input gate, C t , They represent the characteristic variables of the input gating structure at time t respectively; * represents the Hadamard product;

[0053] Output Gate:

[0054] o t =σ(W o ·[h t-1 ,x t ]+b o ) (12)

[0055] h t =o t *tanh(C t ) (13)

[0056] Where o t represents the output of the output gate, W o represents the weight matrix of the output gate, b o represents the bias term of the output gate, h t represents the hidden variable at time t.

[0057] In step 6, set the temperature threshold T of the moving contact threshold ,When the predicted values ​​of the moving contact temperatures of phases A, B, and C obtained by the BiLSTM model exceed the set temperature threshold, an alarm is issued.

[0058] Preferably, the temperature threshold is dynamically adjusted according to different operating scenarios of the switch cabinet, so as to quickly detect safety risks and issue an alarm, including:

[0059] When the ratio of the real-time collected contact pressure value to the rated pressure value is greater than or equal to 1.2, let T threshold =0.8T base , where T base Indicates the reference value of the moving contact temperature threshold;

[0060] When the ratio of the real-time collected temperature value on the top or back of the switch cabinet to the standard temperature value is greater than or equal to 1.1, let T threshold =0.75T base .

[0061] The central server combines the ultrasonic auxiliary diagnosis system to compare and judge the predicted value of the moving contact temperature of phases A, B, and C with T in real time. threshold The predicted value exceeds T thresholdWhen an alarm occurs, an action signal is sent to the LED alarm light and buzzer of the alarm unit.

[0062] Compared with the prior art, the beneficial effects of the present invention include:

[0063] 1) The present invention realizes real-time detection and alarm of high-voltage switchgear groups, reduces the accident rate of closed high-voltage switchgear, improves the ability of the power system to provide safe and reliable power supply, reduces economic losses, and increases social benefits;

[0064] 2) The present invention uses a completely noise-assisted aggregated empirical mode decomposition method to decompose the time series data of the moving contact temperature of the switch cabinet into multiple intrinsic mode components with different frequency domain characteristics, and then uses a neural network model to establish a nonlinear relationship between the intrinsic mode components of the moving contact temperature and various physical quantities of the switch cabinet, thereby improving the prediction accuracy of the moving contact temperature of the switch cabinet;

[0065] 3) The present invention utilizes the forward and backward learning capabilities of BiLSTM to learn the phenomenon that the physical quantities of the switch cabinet change at different speeds when the switch cabinet is overheated, captures the change law of the physical quantities, and proactively perceives the upcoming safety risks such as overheating of the switch cabinet, discovers the safety risks of the switch cabinet operation as early as possible, and issues an alarm so that the operation and maintenance personnel can take timely measures to ensure power safety;

[0066] 4) The present invention uses the Spearman coefficient to analyze the correlation between the physical quantities of the switch cabinet and the intrinsic modal components of the moving contact temperature of each phase. Compared with the Pearson coefficient and the Kendall coefficient, the Spearman coefficient is less susceptible to abnormal data interference and has lower requirements on data distribution. It is particularly suitable for correlation calculation of 10kV switch cabinet data;

[0067] 5) The present invention adopts a dynamically adjusted temperature threshold, which helps to quickly discover risks and issue alarms;

[0068] 6) The present invention adopts an ultrasonic auxiliary diagnosis system to realize the judgment and location detection of partial discharge faults in 10kV switch cabinets, which helps to quickly discover risks and issue alarms;

[0069] 7) The MySQL database used in the present invention realizes the storage and query of data with typical relational characteristics in the switch cabinet holographic data, and has the advantages of small size, low cost, fast computing speed, safety and reliability; the Redis system realizes the efficient storage and query of non-relational data in the switch cabinet holographic data. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0071] Figure 1 Schematic diagram of a switch cabinet monitoring system according to an embodiment of the present invention.

[0072] Figure 2 This is a schematic diagram of the distribution of wireless sensors for collecting holographic data of a switch cabinet according to an embodiment of the present invention.

[0073] Figure 3 It is a schematic diagram of the structure of the central server of an embodiment of the present invention.

[0074] Figure 4 Schematic diagram of the intrinsic mode components obtained by CEEMDAN decomposition in an embodiment of the present invention.

[0075] Figure 5 Schematic diagram of a BiLSTM model according to an embodiment of the present invention.

[0076] Figure 6 Schematic diagram of an ultrasound-assisted diagnosis system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0077] The embodiment relates to safety monitoring of a 10 kV switchgear group of a substation with a voltage level of 220 kV or above.

[0078] like Figure 1-3 As shown, the switch cabinet monitoring system of the embodiment includes a wireless measuring unit and a central server. The wireless measuring unit monitors the real-time temperature value and real-time pressure value of the high-voltage switch cabinet. The wireless measuring unit includes a wireless temperature sensor, a wireless pressure sensor, and an ultrasonic sensor.

[0079] The data transmission network unit receives real-time data measured by wireless sensors through the ZigBee wireless network, and then transmits the real-time data to the central server for data preprocessing and prediction. When the high-voltage switchgear is about to overheat during operation, the switchgear monitoring system will issue an alarm.

[0080] The central server unit uses a completely noise-assisted aggregated empirical mode decomposition method to decompose the moving contact time data series data into intrinsic mode components with different frequency scales, and then uses the Spearman coefficient to select appropriate historical features to establish a bidirectional long short-term memory neural network prediction model, and combines the high-voltage switchgear operating procedures and other physical quantities obtained by measurements related to the moving contact temperature to determine the risk threshold of the high-voltage switchgear in different scenarios; the predicted data is compared with the alarm threshold, and if the warning value is exceeded, the alarm signal is transmitted to the LED alarm light and buzzer of the alarm module for corresponding action.

[0081] The wireless sensor group arranged on the switch cabinet in the embodiment is as follows Figure 2As shown in the figure, 8 wireless temperature sensors are installed at the moving contacts of the A, B, and C three-phase inlet ports in the high-voltage switch cabinet, the monitoring points on the outer surface of the high-voltage switch cabinet, and the static contacts in the switch cabinet. 3 wireless pressure sensors are installed at the contacts in the contact box, and 4 wireless ultrasonic sensors are installed at different positions on the surface of the switch cabinet. The ZigBee wireless network module receives the data sent by each wireless sensor; the ZigBee wireless network module selects the ZigBee platform SoC chip MC13213; the data transmitted by the wireless sensor is received by the ZigBee to IP / TCP device, and the data received by the device is transmitted to the central server via Ethernet. The staff can quickly access the server data through the computer or mobile phone, thereby realizing remote monitoring of the operation status of the high-voltage switch cabinet; the real-time data of the wireless sensor node is transmitted to the ZigBee wireless network module via the RS485 bus interface, and then the real-time data is transmitted to the central server through the coordinator and the RS485 interface converter.

[0082] Central server such as Figure 3 As shown in the figure, since the real-time data collected by the wireless sensor has non-stationary and nonlinear characteristics, the central server stores, preprocesses and nonlinearly predicts the received real-time data, and uses the completely noise-assisted aggregated empirical mode decomposition (CEEMDAN) to decompose the moving contact temperature and ultrasonic signal of the A, B, and C phase incoming ports collected within the time interval, and decomposes the data into intrinsic mode components (IMFs) with different frequency domain characteristics.

[0083] The switch cabinet safety monitoring method based on holographic data combined with ultrasonic signal diagnosis of the embodiment includes the following steps:

[0084] Step 1: Collect the time series data of the temperature and pressure of the switch cabinet and the ultrasonic signal of the arc or corona phenomenon of the shell to obtain the holographic data of the switch cabinet;

[0085] Use temperature sensors to collect the temperature of the moving contacts of phases A, B, and C of the switch cabinet respectively; collect the temperature of the static contacts of phases A, B, and C and the temperature of the top and back of the switch cabinet respectively; use pressure sensors to monitor the pressure at the contacts of phases A, B, and C in the contact box; use four ultrasonic sensors to detect ultrasonic signals of arc or corona phenomena in the switch cabinet housing;

[0086] Step 2: The time series data of the moving contact temperature of phases A, B, and C and the ultrasonic signal are decomposed into multiple intrinsic mode components using the fully noise-assisted aggregate empirical mode decomposition method;

[0087] Step 3: Taking the eigenmode components of the temperature of each phase moving contact of the switch cabinet as the target variable, the Spearman correlation analysis is performed on the holographic data of the switch cabinet to calculate the correlation between each physical quantity of the switch cabinet and the eigenmode components of the temperature of each phase moving contact; the correlation coefficients of the eigenmode components of the ultrasonic signal and the original signal are calculated respectively;

[0088] Step 4: According to the correlation with the moving contact temperature, the main related physical quantities of the moving contact temperature of each phase are screened out; according to the correlation coefficient with the original signal, the intrinsic mode component of the fault ultrasonic signal is screened out;

[0089] Step 5: Establish an ultrasonic signal-assisted diagnosis system. Use the intrinsic mode components of the fault ultrasonic signal selected in step 4 to calculate the wavelet packet energy characteristic spectrum. Then, determine whether a partial discharge fault occurs in the switch cabinet based on the ultrasonic signal characteristics. The ultrasonic signal-assisted diagnosis system is as follows: Figure 6 As shown;

[0090] Step 5.1: Determine whether the intrinsic mode component contains a mutation signal. If it contains a mutation signal, calculate the correlation coefficient between the ultrasonic intrinsic mode component and the original signal to filter out the faulty ultrasonic signal;

[0091] Step 5.2: Calculate the wavelet packet energy characteristic spectrum of the fault ultrasonic signal respectively, determine whether partial discharge occurs in the switch cabinet based on the characteristics of the energy characteristic spectrum, and calculate the location of the partial discharge fault based on the specific location of the ultrasonic sensor and the time difference of the fault ultrasonic signals received by different ultrasonic sensors;

[0092] Taking the ultrasonic sensor that first receives the discharge fault ultrasonic signal as the reference, the time t at which the sth ultrasonic sensor receives the discharge fault ultrasonic signal is s The time difference t from the reception time t1 of the reference ultrasonic sensor s1 =t s -t1, satisfies the following distance equation:

[0093]

[0094] Where v is the propagation speed of the ultrasonic signal, (x, y, z) is the coordinate of the discharge fault point; (x1, y1, z1) is the position coordinate of the reference ultrasonic sensor; (x s ,y s ,z s ) represents the position coordinates of the sth ultrasonic sensor;

[0095] The coordinates (x, y, z) of the discharge fault point can be obtained by solving the distance equation, that is, the discharge fault point can be located in three-dimensional space by using the discharge fault ultrasonic signals received by the reference ultrasonic sensor and the other three ultrasonic sensors.

[0096] Step 6: Establish a BiLSTM model, use the main relevant physical quantities of the moving contact temperature obtained in step 4 as the input of the BiLSTM model, and use the BiLSTM model to predict the intrinsic mode components of the moving contact temperature of each phase; synthesize the intrinsic mode components of the moving contact temperature obtained by the BiLSTM model to obtain the predicted value of the moving contact temperature;

[0097] Step 7: Issue an early warning of the safety risk of the switch cabinet based on the predicted value of the moving contact temperature of the switch cabinet and ultrasonic-assisted diagnosis.

[0098] Fully noise-assisted aggregated empirical mode decomposition method, including:

[0099] Step 1: Randomly add m pairs of positive and negative white noise to the original data sequence;

[0100] Step 2: Decompose the data sequence with added positive and negative white noise using the empirical mode decomposition algorithm, and perform ensemble average processing on the IMFs obtained after adding different white noise pairs to complete one iteration;

[0101] Step 3: Repeat steps 1 and 2 above, and add white noise to the residual value again in each iteration;

[0102] Step 4: After the iteration conditions are met, the complete CEEMDAN decomposition results are obtained.

[0103] Then, the bidirectional long short-term memory neural network prediction model is used to regress and predict the various IMF components in the decomposition results of CEEMDAN, and finally the predicted values ​​of each IMF component are cumulatively reconstructed to obtain the complete data prediction results. Combined with the high-voltage switchgear operating procedures and other related physical quantities, the risk classification thresholds of high-voltage switchgear in different scenarios are determined. The experimental analysis shows that when the ratio of real-time pressure to rated pressure is 1.2 or above, the alarm threshold is lowered by about 20%; when the ratio of other real-time temperatures other than the moving contact to the rated temperature is 1.1 or above, the alarm threshold is lowered by about 25%.

[0104] In the embodiment, after collecting the holographic data of the switch cabinet, the moving contact temperatures of the A, B, and C phase inlets collected within the time interval are decomposed and processed using a completely noise-assisted aggregation empirical mode decomposition method to decompose the data into eigenmode components with different frequency domain characteristics. The specific process is as follows:

[0105] S1: Add the positive and negative white noise pairs to the original data sequence and use the EMD method to decompose and obtain the first modal component IMF1:

[0106]

[0107] The residual component is:

[0108]

[0109] S2: Add positive and negative paired Gaussian white noise to the residual component r1(t) to obtain a new sequence and perform multiple EMD decompositions to obtain IMF2:

[0110]

[0111] Where E k (·) represents the kth IMF component obtained by the EMD algorithm, η j (t) is the jth Gaussian white noise sequence.

[0112] The residual component is:

[0113]

[0114] S3: Repeat the above steps to decompose and obtain the kth residual component:

[0115]

[0116] S4: Repeat the EMD decomposition operation on the processed sequence to derive the k+1th modal component IMF k+1 :

[0117]

[0118] Repeat the above steps until the residual component is a monotonic function that is not suitable for further decomposition, and the decomposition process ends.

[0119] S5: Assuming that L modal components are obtained, the original sequence can be expressed as:

[0120]

[0121] Where r(t) is the final residual component.

[0122] The decomposed eigenmode components are as follows: Figure 4 As shown, Original represents the original measured temperature series, IMF1-IMF6 are the main modal components obtained by decomposition, and r is the residual component.

[0123] In step 5, the specific process of calculating the wavelet packet energy characteristic spectrum is as follows:

[0124] First, the signal is decomposed by wavelet packet to construct the orthogonal scaling function ω(t) and orthogonal wavelet function

[0125]

[0126]

[0127] Where h(p) and g(p) are the filter coefficients in the multi-scale function;

[0128]

[0129]

[0130] In the formula is the pth coefficient corresponding to the node (l,k) after wavelet packet decomposition, and the node (l,k) represents the kth frequency band of the lth layer; They represent the coefficients of the 2kth and 2k+1th frequency bands of the l+1th layer respectively;

[0131] The wavelet packet coefficients are reconstructed as:

[0132]

[0133] According to Parseval's identity:

[0134]

[0135] Where f(x) represents the time domain signal; d(l,p) represents the pth coefficient corresponding to the lth layer;

[0136] The energy E corresponding to the mth frequency band of the lth layer m,l The calculation formula is as follows:

[0137]

[0138] Where M is the length of the mth frequency band; d m,l (p) represents the p-th coefficient of the m-th frequency band of the l-th layer;

[0139] Extract features from the wavelet energy spectrum composed of the energy of all nodes and normalize them to obtain the fault characteristic value Percentage_E l :

[0140]

[0141] The fault type is determined based on the size of the fault characteristic value. If the fault characteristic values ​​of different frequency bands are relatively average, no fault has occurred; if the fault characteristic values ​​of certain frequency bands vary greatly, it is determined that a partial discharge fault has occurred.

[0142] For the data sequence of the intrinsic mode components that have been decomposed and processed, the Spearman coefficient is used to select the appropriate historical characteristics of the physical quantity of the switch cabinet to construct a bidirectional long short-term memory neural network prediction model to realize the moving contact temperature prediction. The specific process is as follows:

[0143] 1) Using the intrinsic mode components IMF1-IMF6 and residual component r obtained by CEEMDAN decomposition, combined with the Spearman coefficient, appropriate historical features are selected to construct a BiLSTM neural network model, and learning and training are performed to make f(x) as close to the original sample as possible, that is, the error is closer to zero:

[0144] f(x)=ω T x+b (8)

[0145] where d i is the order difference after the variables are sorted, and ω and b are the parameters of the training model.

[0146] 2) Set the error calculation interval (ε - ,ε + ), when the error value is greater than the error calculation interval, the error loss value is calculated.

[0147] The proposed BiLSTM prediction model is expressed as follows:

[0148] Based on RNN, LSTM improves on problems such as gradient disappearance, and uses unique gating structures such as input gate, forget gate and output gate to control the transmission of information to cell units through information selection method.

[0149] Forget Gate:

[0150] f t =σ(W f ·[h t-1 ,x t ]+b f ) (9)

[0151] The forget gate reads h t-1 and x t Output a value between 0 and 1 to determine whether the information is transmitted, where 0 means that the information is completely discarded, and 1 means that the information is completely retained and transmitted.

[0152] Input Gate:

[0153] i t =σ(W i ·[h t-1 ,x t ]+b i ) (10)

[0154]

[0155] Output Gate:

[0156] o t =σ(W o ·[h t-1 ,x t ]+b o ) (12)

[0157] h t =o t *tanh(C t ) (13)

[0158] In the formula, f t 、i t , C t , o t 、h t represents the characteristic variable of each gating structure at time t; W represents the weight matrix of each gating structure; b represents the bias term; * is the Hadamard product; h t-1 、h t Represents the input at time t-1 and time t.

[0159] 3) BiLSTM adopts a two-layer LSTM structure. The first layer learns historical information, and the second layer transmits information in the reverse direction to consider future data. By training the data in both the positive and negative directions, the model network can fully and effectively learn the information of the load data. Its network structure is as follows: Figure 5 shown.

[0160] 4) The predicted values ​​of each modal component IMF1-IMF7 are accumulated and reconstructed to achieve the final data prediction, and the mean absolute error MAE, root mean square error RMSE, and mean absolute percentage error MAPE are used to evaluate the prediction model.

[0161] The calculation method of the Spearman coefficient of the embodiment refers to the calculation method of the Spearman coefficient disclosed in the paper "Electric Vehicle Charging Load Interval Prediction Based on Multi-Correlated Daily Scenarios Generation" by Huang Nantian et al. published in the 41st issue of "Proceedings of the Chinese Society of Electrical Engineering" in 2021.

[0162] The bidirectional long short-term memory neural network prediction model provides high-quality data support for effective prediction and alarm. According to the operation requirements of 10kV high-voltage switchgear and the operation procedures of high-voltage equipment in the power system, the risk classification threshold of the closed high-voltage switchgear in different scenarios is determined. According to the pressure data that has been obtained, the test analysis shows that when the ratio of real-time pressure to rated pressure is 1.2 or above, the alarm threshold is lowered by about 20%; when the ratio of other real-time temperatures other than the moving contact to the rated temperature is 1.1 or above, the alarm threshold is lowered by about 25%. When the relevant data of the high-voltage switchgear exceeds the alarm threshold, the relevant alarm equipment will act: the LED alarm light continues to flash red and the buzzer starts, and the display terminal displays a fault. The central server can display the real-time temperature, pressure, ultrasonic signal and fault information of each monitoring point of the high-voltage switchgear, and send corresponding alarm information, so that the operation and maintenance personnel can check the relevant operation status of the switchgear in time, handle related faults, eliminate potential risks as soon as possible, and avoid further expansion of losses.

[0163] The preferred implementation cases of the present invention are specifically described and introduced in combination with the above explanatory text, but the present invention is not limited in any form. Those skilled in the art should understand that, based on the technical essence of the present invention, within the spirit and principles of the present invention, the above implementation cases do not involve creative work, and any simple modifications, changes, improvements and other alternatives all fall within the scope of protection of the technical solution of the present invention and the claims.

Claims

1. A switch cabinet safety monitoring method based on holographic data combined with ultrasonic signal diagnosis, characterized in that: The following steps are involved: Step 1: Collect the time series data of the temperature and pressure of the switch cabinet and the ultrasonic signal of the arc or corona phenomenon of the shell to obtain the holographic data of the switch cabinet; Collect the temperature of the moving contacts of the A, B, and C phases of the switch cabinet respectively; collect the temperature of the static contacts of the A, B, and C phases respectively; use pressure sensors to monitor the pressure at the contacts of the A, B, and C phases in the contact box; use multiple ultrasonic sensors to detect the ultrasonic signals of arc or corona phenomena in the switch cabinet housing; Step 2: The time series data of the moving contact temperature of phases A, B, and C and the ultrasonic signal are decomposed into multiple intrinsic mode components using the fully noise-assisted aggregate empirical mode decomposition method; Step 3: Taking the eigenmode components of the temperature of each phase moving contact of the switch cabinet as the target variable, the Spearman correlation analysis is performed on the holographic data of the switch cabinet to calculate the correlation between each physical quantity of the switch cabinet and the eigenmode components of the temperature of each phase moving contact; calculate the correlation coefficient between the eigenmode components of the ultrasonic signal of the arc or corona phenomenon of the shell and the ultrasonic signal of the switch cabinet under normal working conditions; Step 4: According to the correlation with the moving contact temperature, the main related physical quantities of the moving contact temperature of each phase are screened out; according to the correlation coefficient between the eigenmode components of the ultrasonic signal, the eigenmode components of the fault ultrasonic signal are screened out; Step 5: Establish an ultrasonic signal auxiliary diagnosis system, calculate the wavelet packet energy characteristic spectrum of the fault ultrasonic signal eigenmode component screened out in step 4, and then determine whether a partial discharge fault occurs in the switch cabinet based on the ultrasonic signal characteristics; Step 6: Build and use the BiLSTM model to obtain the predicted value of the moving contact temperature. The specific process includes: 1) The intrinsic mode components IMF1-IMF6 and residual components of the time series of the moving contact temperature of phases A, B, and C obtained in step 2 are respectively r As the prediction variable of the BiLSTM model, the main relevant physical quantities of the moving contact temperature selected in step 4 are used as the input of the BiLSTM model to build the BiLSTM model; Predictors for the BiLSTM model: ；（8） In the formula represents the weight matrix, b represents the bias term, Represents input; The collected holographic data of the switch cabinet is used as the training data set of the BiLSTM model for training the BiLSTM model; 2) Set the error calculation interval ,in , are the lower and upper limits of the error value, respectively. When the error value is greater than the error calculation interval, the error loss value is calculated; 3) In the first LSTM layer of the BiLSTM model, the error is forward propagated to learn from historical data; in the second LSTM layer, the error is backward propagated to learn from future data; the BiLSTM model is trained in both positive and negative directions using the holographic data of the switchgear, so that the BiLSTM model can fully and effectively learn the load information in the holographic data of the switchgear; 4) According to the prediction variables of the trained BiLSTM model, the intrinsic mode components IMF1-IMF6 and residual components of the time series of the moving contact temperature of phases A, B, and C are obtained. r , and synthesize the predicted values ​​of the moving contact temperatures of phases A, B, and C; the mean absolute error MAE, root mean square error RMSE, and mean absolute error percentage MAPE are used to evaluate the prediction accuracy of the BiLSTM model; Step 7: Issue an early warning of the safety risk of the switch cabinet based on the predicted value of the moving contact temperature of the switch cabinet.

2. The switch cabinet safety monitoring method according to claim 1, characterized in that: In step 2, the time series data of the moving contact temperature of the A, B, and C phase inlets collected within the time interval and the ultrasonic signal of the arc or corona phenomenon of the shell are decomposed and processed using the complete noise-assisted aggregation empirical mode decomposition method to obtain intrinsic mode components with different frequency domain characteristics. The specific process includes: S1: Add positive and negative white noise pairs to the original data sequence In the example, the empirical mode decomposition method is used to decompose the first modal component : ；（1） The residual component is: ；（2） In the formula N Indicates the length of the sequence; represents the first residual component; S2: In the residual component Add positive and negative paired Gaussian white noise to obtain a new sequence, and perform empirical mode decomposition to obtain the second modal component : ；（3） In the formula represents the first k IMF components, For the Gaussian white noise sequence; represents the noise factor; The residual component is: ；（4） In the formula represents the second residual component; S3: Similar to step S2, decompose to obtain Residual components: ；（5） In the formula , Respectively represent k -1. k Residual components; Repeatedly use the empirical mode decomposition to decompose the sequence to be processed and derive the modal components : ；（6） S4: repeat step S3 until the residual component obtained is a monotonic function that is not suitable for further decomposition, and then end the decomposition process; Assume the number of modal components obtained is L , then the original sequence can be expressed as: ；（7） In the formula is the final residual component.

3. The switch cabinet safety monitoring method according to claim 2, characterized in that: The specific process of judging whether a partial discharge fault occurs in the switch cabinet based on the characteristics of the ultrasonic signal is as follows: 1) Determine whether the eigenmode component contains a mutation signal. If it does, calculate the correlation coefficient between the ultrasonic eigenmode component and the ultrasonic signal of the switch cabinet under normal working conditions to screen out the fault ultrasonic signal; 2) Calculate the wavelet packet energy characteristic spectrum of the fault ultrasonic signal respectively, judge whether partial discharge occurs in the switch cabinet based on the characteristics of the energy characteristic spectrum, and calculate the location where the partial discharge occurs based on the specific location of the ultrasonic sensor and the time difference between the fault ultrasonic signals received by different ultrasonic sensors.

4. The switch cabinet safety monitoring method according to claim 3, characterized in that: In step 5, the specific process of calculating the wavelet packet energy characteristic spectrum is as follows: First, the signal is decomposed by wavelet packet to construct an orthogonal scaling function and orthogonal wavelet functions : ； ； In the formula These are the filter coefficients in the multi-scale function; ； ； In the formula Node after wavelet packet decomposition The corresponding coefficients, nodes Indicates Layer frequency bands; , Respectively represent l +1 level 2 k , 2 k +1 coefficient of the band; The wavelet packet coefficients are reconstructed as: ； According to Parseval's identity: ； In the formula represents a time domain signal; Indicates The frequency band of the layer coefficients; No. Layer The energy corresponding to the frequency band The calculation formula is as follows: ； In the formula, For the The length of the frequency band; Indicates Layer The frequency band corresponds to coefficients; Extract features from the wavelet energy spectrum composed of the energy of all nodes and normalize them to obtain the fault feature value : 。 5. The switch cabinet safety monitoring method according to claim 4, characterized in that: The first and second LSTM layers of the BiLSTM model use the following gating structure to control the transmission of information to the cell units through the information selection method: Forget Gate: ； （9） In the formula represents the output of the forget gate, represents the weight matrix of the forget gate, represents the activation function; Represents the bias term of the forget gate; express t -1 hidden variables at time; express t Input of time; The forget gate reads and , outputs a value of 0 to 1 to determine whether the information is transmitted. An output value of 0 indicates that the information is completely discarded, and an output value of 1 indicates that the information is completely retained and transmitted; Input Gate: ；（10） ；（11） In the formula represents the output of the input gate, represents the weight matrix of the input gate, represents the bias term of the input gate, , Represent the state at time t and the current input state respectively; represents the Hadamard product; Output Gate: ；（12） ； （13） In the formula represents the output of the output gate, represents the weight matrix of the output gate, represents the bias term of the output gate, express t Hidden variables at time.

6. The switch cabinet safety monitoring method according to claim 5, characterized in that: In step 7, set the temperature threshold of the moving contact T threshold ,When the predicted values ​​of the moving contact temperatures of phases A, B, and C obtained by the BiLSTM model exceed the set temperature threshold, an alarm is issued; According to different operating scenarios of the switch cabinet, the temperature threshold is dynamically adjusted to quickly detect safety risks and issue alarms, including: When the ratio of the real-time collected contact pressure value to the rated pressure value is greater than or equal to 1.2, T threshold =0.8 T base ,in T base Indicates the reference value of the moving contact temperature threshold; When the ratio of the real-time collected temperature value on the top or back of the switch cabinet to the standard temperature value is greater than or equal to 1.1, T threshold =0.75 T base .

Images