Intelligent modulation method for abnormal current of high and low voltage power distribution cabinet

Abstract

The invention relates to the technical field of high and low voltage, and discloses a high and low voltage power distribution cabinet abnormal current intelligent modulation method, which comprises the following steps: respectively aiming at three phase sequences, obtaining real-time current sequences at multiple moments, analyzing current quality based on the real-time current sequences obtained by sampling, constructing a real-time current matrix according to the real-time current sequences, and calculating the abnormal current of a high and low voltage power distribution cabinet according to the real-time current matrix. And the current matrix is compared with a standard current matrix, so that whether the current is abnormal or not can be quickly and directly judged. According to the method, a standard current matrix and a real-time current matrix are differenced, and an abnormal current sample combination capable of forming an overall difference current matrix is searched in an abnormal current sample set according to a superposition principle, so that rapid abnormal current analysis can be realized, various abnormal current components can be extracted, and classification and modulation of the abnormal current can be realized. And meanwhile, a control signal is dynamically adjusted through an adaptive fuzzy control method, the abnormal current is dynamically modulated, the control signal is dynamically adjusted according to the modulation effect, and closed-loop control of the modulation process is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of high-low voltage power distribution cabinet, and particularly relates to an intelligent modulation method for abnormal current of high-low voltage power distribution cabinet. BACKGROUND

[0002] The high-low voltage power distribution cabinet is a power distribution device that distributes busbar circuit to each branch load. The current quality has an important influence on the normal operation and service life of the high-low voltage power distribution cabinet. The current waveform in the power grid should be a standard sine wave in the normal state. However, in the actual power grid, due to various interference factors, such as load, power transmission line fault and the like, the current waveform in the power grid is distorted, and the instantaneous value of the current may exceed the carrying capacity of the high-low voltage power distribution cabinet, causing damage to it.

[0003] The reasons for the abnormal deviation of the current from the sine waveform are complex and diverse. Not only the load at the user end will affect the current, but also the state of the high-low voltage power distribution cabinet itself will have an important influence on the current. For example, there may be high-power nonlinear loads at the user end, and the start-stop of these high-power nonlinear loads will generate harmonics, causing distortion of the current waveform. In the field of civil buildings, three-phase imbalance generated at the user end is also a common cause of current abnormalities. At the same time, the high-low voltage power distribution cabinet will also cause abnormal fluctuations in the current due to the aging of internal components (such as changes in contact resistance) after long-term use. Since there are many users connected to the high-low voltage power distribution cabinet, these factors that cause abnormal current may occur simultaneously, and under the combined action of these factors, the waveform of the current changes frequently and diversely. The traditional method for identifying abnormal current is to identify a specific type separately or to determine whether the current is abnormal by threshold judgment. This method not only has a high misjudgment rate, but also cannot accurately identify and classify the abnormal current under the combined action of multiple factors, so that effective measures cannot be taken for modulation. SUMMARY

[0004] In order to solve the technical problem that the prior art cannot analyze the causes of current abnormalities, the present application provides an intelligent modulation method for abnormal current of high-low voltage power distribution cabinet, comprising the following steps: Respectively for three phase sequences, real-time current sequences at multiple time points are obtained, and real-time current matrices of the three phase sequences are respectively constructed, wherein the real-time current sequence includes a busbar current value and a branch current value of each branch connected to the busbar; One phase sequence is selected one by one, and a standard current matrix corresponding to the standard sine waveform is reconstructed according to the real-time current matrix; The real-time current matrix of the phase sequence is subtracted from the standard current matrix of the phase sequence to obtain a difference current matrix; When the F-norm of the difference current matrix of any one phase sequence is greater than the abnormal early warning value, the overall difference current matrix is constructed according to the difference current matrices of the three phase sequences; The abnormal current sample combination matching the overall difference current matrix is searched from the abnormal current sample set, if the matching abnormal current sample combination exists, the abnormal current component is extracted according to the abnormal current sample combination, and the abnormal current component is modulated, if the matching abnormal current sample combination does not exist, the real-time current matrix is manually labeled and stored in the abnormal current sample set; The abnormal current sample set comprises abnormal current samples generated under the action of a plurality of interference factors.

[0005] Technical effects and advantages of the present application: based on the real-time current sequence, the current quality is analyzed, the real-time current matrix is constructed according to the real-time current sequence, and compared with the standard current matrix, whether the current is abnormal can be quickly and directly judged. By subtracting the standard current matrix from the real-time current matrix, according to the superposition principle, the abnormal current sample combination capable of constructing the difference current matrix is searched in the abnormal current sample set, the abnormal current analysis can be realized quickly, and the abnormal current components of various types are extracted, and the classification and modulation of the abnormal current are realized. At the same time, the self-adaptive fuzzy control method is used to dynamically adjust the control signal, and the abnormal current is dynamically modulated, and the control signal is dynamically adjusted according to the modulation effect, and the closed-loop control of current modulation is realized. BRIEF DESCRIPTION OF DRAWINGS

[0006] Figure 1 It is the overall flowchart of the present application.

[0007] Figure 2 It is the acquisition step flowchart of the abnormal early warning value in the present application.

[0008] Figure 3 It is the flowchart of the self-adaptive fuzzy control harmonic suppression in the present application. DETAILED DESCRIPTION

[0009] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0010] In an ideal circuit, all loads are resistive loads, and the change in load size only affects the amplitude of the current. Thus, when the load in the line is determined, the current waveform in the circuit should also be determined, and even considering the influence of loss and measurement error, the deviation of the current waveform from the standard waveform should be within a small range. However, due to the existence of inductive load in the actual circuit, the inductive load will cause the current waveform to change not only in amplitude but also in period. According to Kirchhoff's law, the current at a node in the circuit is equal to the sum of the currents of each branch, which can be understood as the influence of each load on the current in accordance with the principle of superposition.

[0011] In order to protect each branch and the busbar device, a circuit breaker and an air switch are usually arranged on each branch and the busbar of the high-low voltage power distribution cabinet, and a large number of voltmeters and ammeters are also arranged to monitor the voltage and current of each line.

[0012] Based on the above analysis, the present application provides an intelligent abnormal current modulation method for a high-low voltage power distribution cabinet, which refers to Figure 1 , comprising the following steps: S1, for each of the three phase sequences, obtaining a plurality of real-time current sequences at different times, and constructing a real-time current matrix for each of the three phase sequences, wherein the real-time current sequence includes a busbar current and a branch current of each branch connected to the busbar; Taking phase sequence A as an example, the real-time current matrix constructed according to the real-time current sequence is as follows: wherein, is a real-time busbar current vector of phase sequence A, each component representing a current sampling value at a corresponding time, and the form is as follows: is a real-time branch current vector of the nth branch of phase sequence A, and the form and the meaning represented by each component are the same as those of the real-time busbar current vector; The real-time current matrix constructed in the above manner can not only reflect the busbar current state of phase sequence A, but also reflect the current state of each branch of phase sequence A.

[0013] The forms of the real-time current matrices of phase sequence B and phase sequence C are the same as that of phase sequence A.

[0014] Similarly, for the voltage data (i.e., real-time voltage sequence) of the three phase sequences collected during daily operation, a real-time voltage matrix can also be constructed in the above form so as to participate in the calculation in the subsequent abnormal current identification.

[0015] S2, selecting one phase sequence at a time, and reconstructing a corresponding standard current matrix according to the real-time current matrix and the standard sinusoidal waveform; Alternating current exists active power and reactive power, active power represents actual power consumed by load, active power is the product of effective current, effective voltage and power factor. Effective current is usually selected as the average value of all current sampling values in real-time current sequence. Therefore, the standard current waveform under this load can be deduced by this average value, that is, the standard current matrix, the value of each component of the standard current matrix is the effective current multiplied by the trigonometric function of the phase corresponding to the time of the component.

[0016] S3, subtract the real-time current matrix of the phase sequence from the standard current matrix to obtain a difference current matrix; S4, when the F-norm of the difference current matrix of any one phase sequence is greater than an abnormal early warning value, constructing an overall difference current matrix according to the difference current matrices of the three phase sequences; By subtracting the real-time current matrix from the standard current matrix, the difference current matrix reflecting the difference between the current waveform and the standard sinusoidal waveform can be obtained. The identification program is started only when the F-norm of the difference current matrix reaches the abnormal early warning value, thereby avoiding unnecessary operation, reducing the running time of the whole system and reducing the energy consumption of the monitoring system.

[0017] The abnormal early warning value affects the sensitivity of the control system to abnormal current. The setting of the abnormal early warning value needs to be dynamically adjusted according to the capacity of the high-low voltage power distribution cabinet, the effective current of the real-time current sequence (representing the load rate of the high-low voltage power distribution cabinet at the current time) and the electricity peak and valley. During the electricity peak period, because the high-low voltage power distribution cabinet is in a high load running state for a long time, its anti-impact ability to abnormal current fluctuation decreases, so the detection of abnormal current should be more sensitive. During the electricity valley period, the abnormal early warning value can be appropriately increased to avoid frequent adjustment of the current, on the one hand, the operation time of the control system can be reduced, and on the other hand, overcompensation can be effectively prevented.

[0018] Specifically, the abnormal early warning value can be dynamically adjusted by the following steps: S41, in step S1, the real-time voltage sequence of the three phase sequences is obtained at the same time as the real-time current sequence of the three phase sequences is obtained; S42, for each phase sequence, the active power is calculated according to the real-time voltage sequence and the real-time current sequence; S43, the load rate of the high-low voltage power distribution cabinet is calculated according to the active power, the base early warning value is corrected according to the load rate and the load sensitivity coefficient to obtain the abnormal early warning value Thre_adjt.

[0019] The base early warning value Thre_base can be corrected to obtain the abnormal early warning value Thre_adjt by the following formula: Thre_adjt = Thre_base * (1 + k1 * (1 - Load_ratio)) In the formula, k1 is a load-sensitive coefficient, and Load_ratio is a load ratio of the high-low voltage power distribution cabinet.

[0020] The load-sensitive coefficient k1 is used to reflect the influence of the load ratio of the high-low voltage power distribution cabinet on the abnormal early warning value, and is related to the internal temperature, the internal humidity, and the external environment temperature of the high-low voltage power distribution cabinet. The specific value can be directly set according to prior experience, for example, k1 = 0.2-0.4 according to the current use experience.

[0021] In the case where there is no enough experience to guide the setting, the application further provides a method for automatically adjusting the load-sensitive coefficient k1, which specifically includes the following steps: S431, obtaining the real-time internal temperature, the real-time internal humidity, the real-time environment temperature of the high-low voltage power distribution cabinet, and the real-time voltage sequence of the three phase sequences; S432, inputting the real-time current sequence, the real-time voltage sequence of the three phase sequences, and the real-time internal temperature and the real-time internal humidity of the high-low voltage power distribution cabinet into the digital twin model of the high-low voltage power distribution cabinet, simulating the internal temperature change of the high-low voltage power distribution cabinet, and outputting the simulated internal temperature; S433, dynamically adjusting the load-sensitive coefficient, so that the simulated internal temperature is lower than the temperature upper limit within the adjustment period, and outputting the load-sensitive coefficient.

[0022] Since the internal temperature significantly affects the service life of the high-low voltage power distribution cabinet, using the internal temperature as a limiting condition to control the load-sensitive coefficient can fully tap the carrying capacity of the high-low voltage power distribution cabinet while ensuring the service life.

[0023] After obtaining the overall difference current matrix constructed by the difference current matrices of the three phase sequences, a pre-trained big data model is used to analyze the abnormal current, and a matching abnormal current sample combination is searched from the abnormal current sample set: S5, searching for a matching abnormal current sample combination from the abnormal current sample set, if there is a matching abnormal current sample combination, extracting an abnormal current component according to the abnormal current sample combination, and modulating the abnormal current component, if there is not, manually labeling the real-time current matrix, and storing it in the abnormal current sample set.

[0024] The abnormal current sample set contains abnormal current samples under the action of multiple different types of interference factors. The recording form of these abnormal current samples is the same as that of the real-time current sequence, and the abnormal current samples can be obtained through the following steps: S51, high-frequency sampling is performed on the standard current waveform to obtain normal current sequences at multiple time points. The standard current waveform is a stable 50Hz sine wave when the high-low voltage power distribution cabinet is running at a rated load rate. S52, a type-known interference factor is applied to the standard current waveform at multiple different phases to obtain abnormal current sequences when the interference factor acts at different phases. S53, each abnormal current sequence is subtracted from the normal current sequence to construct an abnormal current sample corresponding to the interference factor. The abnormal current sample includes abnormal current sequences generated when the interference factor acts at multiple different phases.

[0025] In order to improve the amount of data contained in the abnormal current sample, the abnormal current sequence and the normal current sequence can be normalized to unify the amplitude of the current and only retain the shape feature of the waveform. When matching with the overall difference current matrix, more data can be compared to improve the matching success rate and improve the classification accuracy.

[0026] The abnormal current sample set can be measured by actual experiment of the high-low voltage power distribution cabinet, or simulated experiment by the digital twin model of the high-low voltage power distribution cabinet. Since the high-low voltage power distribution cabinet is a relatively complex and expensive device, and the high-low voltage power distribution cabinet needs to be customized according to user requirements, the final state of each device is different, and the data measured by experiment may be affected by the state of the high-low voltage power distribution cabinet participating in the experiment, and there is a certain deviation from the standard current. The abnormal current sample set obtained in this way will introduce new errors when used to identify abnormal current. The digital twin model of the high-low voltage power distribution cabinet established by the simulation experiment is an ideal state mathematical model established according to electromagnetism, thermodynamics and other theories. The abnormal current sample obtained in this way is only affected by the applied interference factor, and is more accurate for abnormal current identification.

[0027] There are many types of interference factors, including not only harmonics generated by nonlinear loads, but also current surges generated by sudden access or exit of high-power loads (such as inrush current), overcurrent caused by load short circuit, load overcapacity, and current fluctuation caused by high-low voltage power distribution cabinet failure.

[0028] More accurately, fluctuations caused by instability of the power supply end (energy source) can also be considered, especially with the access of clean energy such as solar and wind energy. These energy sources are obviously affected by weather factors, and their own instability will have a certain impact on the power grid. Even before it is connected to the grid, there are corresponding filtering and voltage stabilizing measures, but it still brings fluctuations to the power grid more or less. Therefore, when establishing the abnormal current sample set, the abnormal current samples generated by these interference factors can be included.

[0029] Specifically, the abnormal current sample combination matching the overall difference current matrix is found by the following steps: S54, normalizing the overall difference current matrix; S55, constructing an abnormal current vector according to each abnormal current sample in the abnormal current sample set, and constructing a matching equation according to all abnormal current vectors and the overall difference current matrix; S56, judging whether the matching equation has a non-zero real solution, and if so, extracting the matched abnormal current sample combination.

[0030] When sampling the current in the circuit, in order to meet the accuracy requirement of analysis, the sampling frequency can reach 10 kHz or even 50 kHz, and these sampling data completely record the fluctuation of the current. After filtering out the noise, the remaining sampling points can fully reflect the characteristics of the current. Therefore, based on the real-time current sequence obtained by sampling, the current quality is analyzed, the real-time standard vector is constructed according to the real-time current sequence, and compared with the standard current vector, it can be quickly and directly judged whether the current is abnormal. By subtracting the standard current matrix from the real-time current matrix, according to the superposition principle, the abnormal current sample combination that can constitute the overall difference current matrix is found in the abnormal current sample set, so that the fast abnormal current analysis can be realized, and the abnormal current components can be extracted, and then the classification and modulation of the abnormal current can be realized.

[0031] After obtaining the abnormal current sample combination matching the overall difference current matrix through the above steps, the abnormal current components can be extracted, and then targeted modulation can be performed.

[0032] The most common factor causing current abnormality in the circuit is the harmonic generated by the inductive load, so when it is judged according to step S5 that there is a harmonic in the difference current matrix, the following steps are performed to manage the harmonic: S561, calculating the real-time THD according to the real-time current sequence, constructing a membership function according to the real-time THD, load change rate and historical distortion mode, and calculating the membership degree of the fuzzy set according to the membership function; Among them, the real-time THD (Total Harmonic Distortion) is obtained by the following steps: S5611, performing frequency domain analysis on the real-time current sequence, extracting the frequency and amplitude of the multi-order harmonic current, calculating the effective value of each order harmonic current, and calculating the effective value of the fundamental wave current; Specifically, it can be calculated according to the following formula: In the formula, is the effective value of the fundamental wave current, is the effective value of the hth harmonic current; S5612, calculating real-time THD according to the effective value of each harmonic current and the effective value of fundamental current; S562, establishing IF-THEN rule base based on expert experience base, and deriving control decision through fuzzy inference engine; S563, converting fuzzy inference result (i.e. control decision) into actual control instruction, such as APF compensation current amplitude, PPF switching combination, etc., to perform harmonic suppression; S564, obtaining real-time current sequence after harmonic suppression, calculating actual THD, and dynamically adjusting membership function parameters and rule weights according to the deviation of actual THD and allowed distortion value (i.e. total harmonic distortion rate allowed in circuit) through deep learning algorithm, so as to continuously optimize control effect.

[0033] Here, the historical distortion mode refers to historical harmonic characteristics. Through k-means clustering algorithm, historical harmonic data is classified and identified, time dimension features (such as seasonal difference, peak and off-peak electricity, etc.), frequency domain features (such as the ratio between each harmonic) and load type features (such as active power, power factor, etc.) in historical harmonic data are extracted, and historical harmonic data with the same features are clustered.

[0034] Adaptive Fuzzy Control (AFC) is an intelligent control method combining fuzzy logic and adaptive algorithm, which can automatically adjust control rules and parameters according to system dynamic changes. In power system harmonic control, it dynamically optimizes filter behavior by responding to real-time THD changes in real time.

[0035] Adaptive Fuzzy Control can not only be applied to harmonic control, but also be applied to the modulation of other types of abnormal currents. For example, when it is identified through step S5 that there is current anomaly caused by inrush current in the circuit, the following steps are performed: S565, calculating real-time deviation and real-time deviation change rate according to real-time current sequence; S566, fuzzy inference engine infers fuzzy value of control output through IF-THEN rule base according to real-time deviation and real-time deviation change rate, and defuzzifies to obtain control signal to suppress inrush current; S567, obtaining real-time current sequence after suppression, calculating actual deviation and actual deviation change rate, and adjusting membership function parameters and rule weights according to actual deviation change rate and the deviation of actual deviation and allowed distortion value, to dynamically adjust control signal and optimize control effect.

[0036] The actual deviation is the deviation between the real-time current sequence after suppression and the real-time current sequence before suppression, and the actual deviation change rate is the change speed of the deviation over time. The calculation form can be realized in the form of vector operation, and the specific form is the same as the real-time current vector, which will not be described here.

[0037] The hardware for realizing harmonic and inrush current suppression is mostly energy storage elements, for example, APF uses IGBT triode to control the charging and discharging of capacitor to realize the governance of harmonic and achieve the purpose of suppressing current distortion. These are all prior art, which will not be described here.

[0038] When there is no matching abnormal current sample combination, it may mean that the high-low voltage power distribution cabinet has an uncommon fault, thereby causing abnormal current. At this time, the overall difference current matrix or the real-time current sequence can be labeled by manual labeling, or the running state of the high-low voltage power distribution cabinet can be analyzed according to other state parameters of the high-low voltage power distribution cabinet, such as internal temperature, internal humidity, ambient temperature, grounding current, etc., to determine whether an abnormality occurs.

[0039] The intelligent management system of the high-low voltage power distribution cabinet will collect a large amount of running data when it is running daily, including the current and voltage of three phase sequences (including bus and each branch), and calculate the active power, reactive power and power factor in each phase sequence according to the current and voltage. Some intelligent management systems of high-low voltage power distribution cabinets will also collect other running state data, such as internal temperature, internal humidity, grounding current, ambient temperature, etc., to comprehensively evaluate the running state of the high-low voltage power distribution cabinet, so as to determine whether the abnormal current is caused by the fault of the high-low voltage power distribution cabinet.

[0040] Specifically, the running state of the high-low voltage power distribution cabinet can be determined by the following steps: S6, obtaining the running state parameters of the high-low voltage power distribution cabinet at the same time as the real-time current sequence, including real-time internal temperature, real-time internal humidity, real-time ambient temperature, etc. S7, constructing a running state matrix of the high-low voltage power distribution cabinet according to the running state parameters and the real-time current sequence and real-time voltage sequence of the three phase sequences; S8, classifying the running state of the high-low voltage power distribution cabinet according to the running state matrix, and determining the fault type of the high-low voltage power distribution cabinet.

[0041] Since the real-time current sequence and the real-time voltage sequence contain a large amount of data in order to maintain the integrity of the current waveform, a large amount of calculation will be generated when identifying the fault type. Therefore, the following steps are used to improve the identification method of the fault type: S9, obtaining the running state parameters of the high-low voltage power distribution cabinet at the same time as the real-time current sequence, including real-time internal temperature, real-time internal humidity, real-time ambient temperature, etc. S10, import the real-time current sequence, real-time voltage sequence and operating state parameters of the three phase sequences into the digital twin model of the high and low voltage power distribution cabinet, simulate the operation of the high and low voltage power distribution cabinet, and obtain the predicted state parameters of the high and low voltage power distribution cabinet; S11, compare the operating state parameters with the predicted state parameters to obtain an abnormal state vector; S12, classify the operating state of the high and low voltage power distribution cabinet according to the abnormal state vector, and determine the fault type of the high and low voltage power distribution cabinet.

[0042] When a certain component of the high and low voltage power distribution cabinet fails, in addition to causing the current of the three phase sequences to be abnormal, it will also cause a certain operating state parameter to deviate from the normal state. Different fault types affect different operating state parameters. The method of comprehensive judgment of multiple parameters can accurately identify the fault type, facilitate maintenance personnel to take effective measures in time, and also can take protective measures in time through existing protection devices. For example, when the bus or branch appears insulation breakdown or interphase short circuit, the quick break protection is carried out in time to avoid the high and low voltage power distribution cabinet from rapidly heating up to cause more serious accidents. Or when the heat dissipation system fails to cause the internal temperature to be too high, the temperature alarm is triggered, the current of the bus or branch is reduced through the cut-in current limiting load or other measures to control the internal temperature and avoid burning out.

[0043] Finally: the above only describes the preferred embodiments of the present application and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A high-low voltage power distribution cabinet abnormal current intelligent modulation method, characterized in that, The method comprises the following steps: obtaining real-time current sequences at multiple time points for three phase sequences respectively, and constructing real-time current matrices for the three phase sequences respectively, wherein the real-time current sequence comprises a bus current value and a branch current value of each branch connected to the bus; selecting one phase sequence at a time, and reconstructing a corresponding standard current matrix according to the real-time current matrix and a standard sinusoidal waveform; subtracting the standard current matrix of the phase sequence from the real-time current matrix of the phase sequence to obtain a difference current matrix; when the F-norm of the difference current matrix of any one phase sequence is greater than an abnormal early warning value, constructing a total difference current matrix according to the difference current matrices of the three phase sequences; finding an abnormal current sample combination matching the total difference current matrix from an abnormal current sample set, if there is a matching abnormal current sample combination, extracting an abnormal current component according to the abnormal current sample combination and modulating the abnormal current component, and if there is not, manually labeling the real-time current matrix and storing it in the abnormal current sample set; the abnormal current sample set comprises abnormal current samples generated under the action of multiple interference factors alone.

2. The method of claim 1, wherein, The abnormal early warning value is obtained through the following steps: simultaneously obtaining real-time voltage sequences of the three phase sequences when obtaining the real-time current sequences of the three phase sequences; for each phase sequence, calculating active power according to the real-time voltage sequence and the real-time current sequence; calculating the load rate of the high-low voltage power distribution cabinet according to the active power, correcting the basic early warning value according to the load rate and a load sensitive coefficient, and obtaining the abnormal early warning value.

3. The method of claim 2, wherein, The load sensitive coefficient is obtained through the following steps: obtaining real-time internal temperature, real-time internal humidity and real-time environmental temperature of the high-low voltage power distribution cabinet; inputting the real-time current sequences, the real-time voltage sequences and the real-time internal temperature, the real-time internal humidity and the real-time environmental temperature of the high-low voltage power distribution cabinet into a digital twin model of the high-low voltage power distribution cabinet to simulate the internal temperature change of the high-low voltage power distribution cabinet, and outputting a simulated internal temperature; dynamically adjusting the load sensitive coefficient to make the simulated internal temperature lower than the temperature upper limit within an adjustment period, and outputting the load sensitive coefficient.

4. The method of claim 1, wherein, Abnormal current samples are obtained through the following steps: high-frequency sampling a standard current waveform to obtain normal current sequences at multiple time points, wherein the standard current waveform is a stable sinusoidal current when the high-low voltage power distribution cabinet is running at a rated load rate; applying a known type of interference factor to the standard current waveform at multiple different phases to obtain abnormal current sequences when the interference factor acts at different phases; subtracting each abnormal current sequence from the normal current sequence to construct an abnormal current sample corresponding to the interference factor, wherein the abnormal current sample comprises abnormal current sequences generated when the interference factor acts at multiple different phases.

5. The method of claim 1, wherein, The abnormal current sample combination matching the total difference current matrix is found through the following steps: normalizing the total difference current matrix; constructing an abnormal current vector according to each abnormal current sample in the abnormal current sample set, and constructing a matching equation according to all abnormal current vectors and the total difference current matrix; determining whether the matching equation has a non-zero real solution, and if so, extracting the matched abnormal current sample combination.

6. The method of claim 1, wherein, When the abnormal current component contains harmonics, the harmonics are treated by the following steps: According to the real-time current sequence, the real-time THD is calculated, the membership function is constructed according to the real-time THD, the load change rate and the historical distortion mode, and the membership degree of the fuzzy set is calculated according to the membership function; Based on the expert experience library, an IF-THEN rule base is established, and a control decision is obtained through a fuzzy reasoning engine; The control decision is converted into an actual control instruction, including the APF compensation current amplitude and the PPF switching combination, for harmonic suppression; After the real-time current sequence after harmonic suppression is obtained, the actual THD is calculated, and through a deep learning algorithm, the deviation between the actual THD and the allowed distortion value is used to dynamically adjust the membership function parameters and rule weights, so that the system continuously optimizes the control effect.

7. The method of claim 6, wherein, The real-time THD is calculated by the following steps: The real-time current sequence is analyzed in the frequency domain, the frequency and amplitude of the multi-order harmonic current are extracted, the effective value of each order harmonic current is calculated, and the effective value of the fundamental wave current is calculated; The real-time THD is calculated according to the effective value of each order harmonic current and the effective value of the fundamental wave current.

8. The method of claim 1, wherein, When the abnormal current component contains inrush current, the inrush current is suppressed by the following steps: According to the real-time current sequence, the real-time deviation and the real-time deviation change rate are calculated; The fuzzy reasoning engine inferences the fuzzy value of the control output through the IF-THEN rule base according to the real-time deviation and the real-time deviation change rate, and de-fuzzifies to obtain the control signal, which suppresses the inrush current; After the real-time current sequence after suppression is obtained, the actual deviation and the actual deviation change rate are calculated, and according to the actual deviation change rate and the deviation between the actual deviation and the allowed distortion value, the membership function parameters and rule weights are adjusted, the control signal is dynamically adjusted, and the control effect is optimized.

9. The method of claim 1, wherein, When there is no abnormal current sample combination in the abnormal current sample set that matches the overall difference current matrix, the following steps are also performed: While obtaining the real-time current sequence, obtain the operating state parameters of the high and low voltage power distribution cabinet, including the real-time internal temperature, the real-time internal humidity and the real-time environment temperature; While obtaining the real-time current sequence, obtain the real-time voltage sequence of the three phase sequences; According to the operating state parameters and the real-time current sequence and real-time voltage sequence of the three phase sequences, the operating state matrix of the high and low voltage power distribution cabinet is constructed; According to the operating state matrix, the operating state of the high and low voltage power distribution cabinet is classified, and the fault type of the high and low voltage power distribution cabinet is determined.

10. The method of claim 1, wherein, When there is no abnormal current sample combination in the abnormal current sample set that matches the overall difference current matrix, the following steps are also performed: While obtaining the real-time current sequence, obtain the operating state parameters of the high and low voltage power distribution cabinet, including the real-time internal temperature, the real-time internal humidity and the real-time environment temperature; While obtaining the real-time current sequence, obtain the real-time voltage sequence of the three phase sequences; The real-time current sequence, the real-time voltage sequence and the running state parameter of three phase sequences are introduced into the digital twin model of the high-low voltage power distribution cabinet, the running of the high-low voltage power distribution cabinet is simulated, and the predicted state parameter of the high-low voltage power distribution cabinet is acquired; The running state parameter and the predicted state parameter are compared, and an abnormal state vector is acquired; The running state of the high-low voltage power distribution cabinet is classified according to the abnormal state vector, and the fault type of the high-low voltage power distribution cabinet is judged.

Images