Automatic fault troubleshooting detection method for electric meter box and electric meter box

Abstract

The present invention relates to the technical field of meter box monitoring, and specifically to an automatic fault troubleshooting detection method for an electric meter box and an electric meter box. The present invention first obtains a multi-dimensional electrical time series data sequence of the electric meter box; further, according to the local fluctuation characteristics of the data, obtains the data anomaly coefficient at each time point, and obtains the abnormal node sequence; further, based on dynamic time warping, obtains node pairs of different abnormal node sequences, and obtains the abnormal factor according to the change characteristics of the mean value of the data anomaly coefficient of the node pair over the time span, combined with the fluctuation characteristics of the mean value of the data anomaly coefficient; finally, according to the change of the abnormal factor, combined with the temperature change in the electric meter box, obtains the current target heat dissipation power and determines whether to issue an early warning, avoids the interference of complex causal relationships caused by inaccurate heat dissipation assessment of the electric meter box and reduces the abnormal performance of multi-dimensional data, reduces the misjudgment of issuing early warning signals, improves the accuracy of early warning, and ensures the stability and safety of equipment operation.

Description

Technical Field

[0001] The invention relates to the technical field of meter box monitoring, and in particular to an automatic fault troubleshooting and detection method for an meter box and an meter box. Background Art

[0002] In the power network, the meter box is not only an indispensable part of the power system, but also undertakes multiple functions such as remote monitoring and management, real-time data processing, fault detection and early warning, so it is necessary to ensure the normal operation of the meter box. Currently, a variety of sensors are used to monitor multi-dimensional electrical data such as temperature, current, voltage, etc. in the meter box, and fault identification and prediction are performed based on time series analysis.

[0003] However, during the automatic troubleshooting process, due to factors such as heat accumulation inside the meter box, changes in the heat dissipation device, and fluctuations in the external environment, it is impossible to accurately distinguish between normal operating fluctuations and potential faults, which can easily lead to misjudgment. At the same time, the abnormal performance of data in different dimensions may not be a strict causal relationship. Traditional methods are difficult to accurately capture the actual fault trend of the equipment and it is difficult to accurately determine whether an early warning signal needs to be issued. Summary of the invention

[0004] In order to solve the technical problem that the existing multi-dimensional data-based monitoring of electric meter boxes is prone to misjudgment, the purpose of the present invention is to provide an automatic fault troubleshooting detection method for electric meter boxes and an electric meter box. The technical solution adopted is as follows:

[0005] A method for automatically troubleshooting and detecting faults of an electric meter box, the method comprising:

[0006] Obtain a multi-dimensional electrical time series data sequence of an electric meter box; select the latest time series data sequence in any dimension as a target sequence; obtain a data anomaly coefficient at each time point according to the local fluctuation characteristics of the data at each time point in the target sequence; obtain an abnormal node sequence in the target sequence based on the data anomaly coefficient;

[0007] Match the latest abnormal node sequences in the temperature dimension and any other dimension through dynamic time warping to obtain node pairs; obtain the fault performance correlation of each node pair according to the change characteristics of the mean value of the data abnormality coefficient of the node pair over the time span;

[0008] In the latest abnormal node sequence in the temperature dimension, the abnormal factor at each time point is obtained according to the fault manifestation correlation of all node pairs corresponding to each time point, combined with the fluctuation characteristics of the mean value of the data abnormal coefficient corresponding to all node pairs; according to the change of the abnormal factor in the latest abnormal node sequence in the temperature dimension, combined with the temperature change in the meter box, the current target heat dissipation power is obtained and it is determined whether to issue an early warning.

[0009] Furthermore, the method for obtaining the data anomaly coefficient includes:

[0010] The derivative of each data on the data curve of the target sequence is obtained; according to the difference characteristics between the derivative of each data and the mean of all the derivatives, combined with the absolute value of the mean of all the derivatives, the data anomaly coefficient of each data corresponding to the time point is obtained; the difference characteristics between the derivative of each data and the mean of all the derivatives are positively correlated with the data anomaly coefficient; the absolute value of the mean of all the derivatives is negatively correlated with the data anomaly coefficient.

[0011] Furthermore, the method for obtaining the abnormal node sequence includes:

[0012] The time point when the data anomaly coefficient is greater than a preset anomaly threshold is marked as an abnormal node; and all the abnormal nodes in the target sequence constitute an abnormal node sequence.

[0013] Furthermore, the method for obtaining the fault manifestation correlation degree includes:

[0014] The mean of the data anomaly coefficient of the node pair is used as the ordinate and the time span is used as the abscissa to form a two-dimensional sample space; the slope of the fitting straight line of all sample points in the two-dimensional sample space is obtained;

[0015] The fault manifestation correlation of each node pair is obtained according to the ratio of the mean value of the data anomaly coefficient of each node pair to the time span and the similarity characteristics of the slope.

[0016] Furthermore, after the ratio of the mean value of the data anomaly coefficient and the time span of each node pair and the absolute value of the difference between the slopes are negatively correlated, the mapping value is used as the fault manifestation correlation degree of each node pair.

[0017] Furthermore, the method for obtaining the abnormal factor includes:

[0018] Taking the mean of the data anomaly coefficients of the node pair as the anomaly parameter of the corresponding node pair;

[0019] In the abnormal node sequence of the latest temperature dimension, the ratio of the mean of the fault performance correlation of all node pairs corresponding to each time point to the variance of the abnormal parameters of all node pairs is used as the abnormal factor of the corresponding time point.

[0020] Furthermore, the method for obtaining the target heat dissipation power includes:

[0021] In the abnormal node sequence with the latest temperature dimension, the product of the abnormal factor with the largest time series and the corresponding temperature data value is used as the first numerator, the product of the abnormal factor with the smallest time series and the corresponding temperature data value is used as the first denominator, and the fractional ratio of the first numerator to the first denominator is used as the adjustment coefficient;

[0022] The product of the adjustment coefficient and the current heat dissipation power is used as the current target heat dissipation power.

[0023] Furthermore, the method for determining whether to issue an early warning includes:

[0024] When the adjustment coefficient is greater than a preset fever threshold, an early warning signal is issued.

[0025] Furthermore, the dimensions of the time series data sequence include at least temperature, humidity, voltage and current; the data of each dimension becomes equal in length after spline interpolation processing.

[0026] The present invention also proposes an electric meter box, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of the automatic fault troubleshooting and detection method for the electric meter box.

[0027] The present invention has the following beneficial effects:

[0028] The present invention first obtains the multi-dimensional electrical time series data sequence of the meter box to obtain the analysis basis; further, according to the local fluctuation characteristics of the data, the data anomaly coefficient of each time point is obtained to quantify the degree of anomaly at the time point to avoid masking important information due to global smoothing, and provide a basis for subsequent analysis; further, based on the data anomaly coefficient, an abnormal node sequence is obtained to facilitate the analysis of the correlation characteristics of anomalies in different dimensions; further, the temperature dimension and the latest abnormal node sequence of any other dimension are matched through dynamic time regularization to obtain node pairs, and the variable corresponding relationship between the abnormal nodes of the two dimensions is extracted to reduce the interference of the complex causal relationship of the abnormal performance of multi-dimensional data, and better reflect the abnormal correlation status in the two dimensions. ; Further obtain the fault performance correlation of each node pair, quantify the strength of the correlation of different node pairs, and provide a basis for subsequent optimization of early warning strategies; further obtain abnormal factors by matching multidimensional data, improve the accuracy of abnormal analysis of temperature dimension, and provide a basis for subsequent adjustment of heat dissipation power and early warning; further reflect the abnormal accumulation trend of heat inside the meter box according to the change of abnormal factors in the latest abnormal node sequence in the temperature dimension, combined with the temperature change in the meter box, adjust the heat dissipation power, and analyze the current abnormal risk of the meter box at the same time, determine whether to warn, avoid misjudgment of issuing early warning signals due to inaccurate heat dissipation assessment of the meter box, improve early warning accuracy, and provide long-distance indication for the work of the meter box in the power grid. The present invention reduces the interference of complex causal relationships of abnormal performance of multidimensional data by extracting abnormal node sequences and matching them through dynamic time regularization, analyzes the abnormal accumulation trend of heat inside the meter box, reduces misjudgment of issuing early warning signals, improves early warning accuracy, and ensures the stability and safety of equipment operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0030] Figure 1 A flow chart of a method for automatically troubleshooting and detecting faults of an electric meter box provided by an embodiment of the present invention;

[0031] Figure 2 A flowchart of a method for acquiring a fault manifestation correlation degree provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the automatic fault troubleshooting detection method and the electric meter box proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0033] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0034] The following describes in detail a method for automatically troubleshooting and detecting faults of an electric meter box and a specific scheme of the electric meter box provided by the present invention in conjunction with the accompanying drawings.

[0035] See also Figure 1 , which shows a flow chart of a method for automatically troubleshooting and detecting faults of an electric meter box provided by an embodiment of the present invention, specifically comprising:

[0036] Step S1: Obtain a multi-dimensional electrical time series data sequence of the electric meter box; select the latest time series data sequence in any dimension as the target sequence; obtain the data anomaly coefficient at each time point according to the local fluctuation characteristics of the data at each time point in the target sequence; obtain the abnormal node sequence in the target sequence based on the data anomaly coefficient.

[0037] In an embodiment of the present invention, multi-dimensional electrical data is acquired by monitoring by setting multiple types of sensors in the electric meter box, and a time series data sequence is formed with a sampling period to obtain a basis for analysis data; the latest time series data sequence in any dimension is selected as the target sequence to facilitate analysis one by one.

[0038] It should be noted that, in one embodiment of the present invention, the dimensions of the time series data sequence include at least temperature, humidity, voltage and current; considering that the data collection frequencies of different dimensions may be different, the data of each dimension becomes equal in length after spline interpolation processing; each collection cycle is set to 10 minutes, the heat dissipation power is adjusted every 10 minutes, and a determination is made as to whether an early warning is issued; the collection frequency can be set by the implementer, and it and the spline interpolation method are both existing technologies and will not be described in detail.

[0039] Taking into account that data anomalies of the meter box usually manifest themselves in the form of local fluctuations, the data anomaly coefficient at each time point is obtained according to the local fluctuation characteristics of the data at each time point in the target sequence. This can accurately capture anomalies, quantify their severity, avoid masking important information due to global smoothing, and provide a reliable basis for subsequent multidimensional correlation analysis, anomaly factor calculation, and early warning mechanism. Further, the abnormal node sequence in the target sequence is obtained based on the data anomaly coefficient, which is convenient for analyzing the correlation characteristics of anomalies in different dimensions.

[0040] Preferably, in one embodiment of the present invention, considering that the curve can more intuitively show the change characteristics of the data, the derivative of the data on the data curve reflects the change trend of the data, so the derivative of each data on the data curve of the target sequence is obtained;

[0041] Considering that the difference between the derivative and the mean of the derivative is large, it means that the difference between the local fluctuation change of the data and the overall fluctuation change is large. At the same time, considering the different fluctuation characteristics of data in different dimensions, the mean absolute value of all derivatives is combined for correction. Based on this, according to the difference characteristics between the derivative of each data and the mean of all derivatives, combined with the absolute value of the mean of all derivatives, the data anomaly coefficient of each data corresponding to the time point is obtained;

[0042] Among them, the difference characteristics between the derivative of each data and the mean of all derivatives are positively correlated with the data anomaly coefficient; the absolute value of the mean of all derivatives is negatively correlated with the data anomaly coefficient.

[0043] As an example, the calculation formula for the data anomaly coefficient includes: ;

[0044] In the formula, i represents the serial number of the data point in the target sequence, which is also the serial number of the time point; Represents the data anomaly coefficient at the i-th time point in the target sequence; represents the hyperbolic tangent function; Represents the derivative corresponding to the i-th time point in the target sequence; Represents the mean of all derivatives corresponding to the target sequence; Indicates taking the absolute value; Indicates preset division by zero positive parameter.

[0045] In the calculation formula of the data anomaly coefficient, the local fluctuation characteristics of the data point are expressed by the derivative, and the difference characteristics between the derivative and the derivative mean are expressed by the absolute value of the difference. The larger it is, the greater the difference between the local fluctuation characteristics of the data and the overall fluctuation characteristics, the more abnormal the corresponding time point is, and the larger the data anomaly coefficient is; The larger it is, the more drastic the overall fluctuation is, the more likely the local fluctuation at a time point is normal fluctuation, and the smaller the data anomaly coefficient is; by presetting the zero-dividing positive parameter, the denominator is prevented from being zero. ; Through the hyperbolic tangent function Partial normalization.

[0046] It should be noted that, in another embodiment of the present invention, considering that the implementer may need to monitor certain data with certain symmetry in the electric meter box, the data anomaly coefficient at the corresponding time point of each data is obtained according to the difference characteristics between the absolute value of the derivative of each data and the mean of the absolute values ​​of all derivatives, combined with the mean of the absolute values ​​of all derivatives, and expressed by the formula: ;

[0047] In the formula, Represents the absolute value of the derivative corresponding to the i-th time point in the target sequence; Represents the mean of the absolute values ​​of all derivatives corresponding to the target sequence.

[0048] The absolute value of the derivative indicates the severity of the local fluctuation of the data. The larger the absolute value of the derivative, the more severe the fluctuation. It shows the difference between the local fluctuation characteristics and the overall fluctuation characteristics of the data, and then uses Indicates the severity of the overall fluctuation. Make corrections.

[0049] It should be noted that converting time series data into a data curve form, obtaining derivatives and hyperbolic tangent functions are all existing technologies and will not be described in detail.

[0050] Preferably, in one embodiment of the present invention, considering that the larger the data anomaly coefficient is, the more likely it is that an abnormal fluctuation will occur at the corresponding time point, so the time point at which the data anomaly coefficient is greater than a preset anomaly threshold is marked as an abnormal node; all abnormal nodes in the target sequence constitute an abnormal node sequence.

[0051] As an example, the preset abnormal threshold is 0.53.

[0052] Step S2: Match the latest abnormal node sequences of the temperature dimension and any other dimension through dynamic time warping to obtain node pairs; obtain the fault performance correlation of each node pair based on the change characteristics of the mean value of the data anomaly coefficient of the node pair over the time span.

[0053] Since the abnormal performances between different dimensions may not have a strict causal relationship, the performance of the same meter box fault condition in the respective dimensional data is not completely consistent. Therefore, the latest abnormal node sequence of the temperature dimension and any other dimension is matched through dynamic time warping to obtain node pairs and extract the variable correspondence between the abnormal nodes of the two dimensions. This reduces the interference of the complex causal relationship of the abnormal performance of multidimensional data, better reflects the abnormal correlation between the two dimensions, and provides a basis for analyzing the abnormal correlation between temperature anomalies and abnormalities in other dimensions.

[0054] Considering that the mean of the data anomaly coefficient of the node pair reflects the overall anomaly intensity, the time span reflects the time interval between abnormal situations, and the changing trend of the mean of the data anomaly coefficient with the time span can reflect the dynamic correlation characteristics between data of different dimensions, the fault performance correlation of each node pair is obtained according to the changing characteristics of the mean of the data anomaly coefficient of the node pair with the time span, and the correlation strength of different node pairs is quantified, providing a basis for the subsequent optimization of the early warning strategy.

[0055] Preferably, in one embodiment of the present invention, the method for obtaining the fault manifestation correlation includes:

[0056] See also Figure 2 , which shows a flow chart of a method for obtaining a fault manifestation correlation degree provided by an embodiment of the present invention, specifically comprising:

[0057] Step S201: Taking the mean of the data anomaly coefficient of the node pair as the ordinate and the time span as the abscissa to form a two-dimensional sample space; obtaining the slope of the fitting straight line of all sample points in the two-dimensional sample space.

[0058] Considering that the more similar the abnormal characteristics of a node pair are to the overall abnormal trend of all node pairs, the more this node pair conforms to the overall variable characteristics and the greater the correlation of fault manifestations, the overall abnormal trend of the node pair is analyzed first: the mean of the data abnormality coefficient of the node pair is taken as the ordinate and the time span is taken as the abscissa to form a two-dimensional sample space; the slope of the fitting straight line of all sample points in the two-dimensional sample space is obtained, and the slope reflects the overall trend of the data abnormality intensity changing with the time span.

[0059] Step S202: Obtain the fault manifestation correlation of each node pair according to the ratio of the mean value of the data anomaly coefficient of each node pair to the time span and the similarity characteristics of the slope.

[0060] As an example, the ratio of the mean value of the data anomaly coefficient of each node pair to the time span and the absolute value of the difference in slope are negatively correlated and the mapped value is used as the fault manifestation correlation of each node pair.

[0061] The calculation formula of fault manifestation correlation includes: ;

[0062] Among them, r represents the serial number of other dimensions except the temperature dimension; s represents the serial number of the node pair; It represents the fault performance correlation of the sth node pair when the temperature dimension is matched with the latest abnormal node sequence of the rth other dimension; It represents the exponential function with the natural constant e as the base; It represents the slope of the fitted line when the temperature dimension is matched with the latest abnormal node sequence of the rth other dimension; It represents the mean value of the data anomaly coefficient of the sth node pair when the temperature dimension is matched with the latest anomaly node sequence of the rth other dimension; Represents the time span of the sth node pair when the temperature dimension is matched with the latest abnormal node sequence of the rth other dimension.

[0063] In the calculation formula of fault manifestation correlation, Reflects the overall trend of data anomaly intensity over time, through It shows the relationship between the overall data anomaly coefficient of a single node pair and the time span, characterizes the abnormal characteristics of the node pair, and expresses the consistency between the abnormal characteristics of the node pair and the overall trend through the absolute value of the difference, reflecting the correlation strength of the node pair. The smaller the difference, the stronger the correlation. The greater the correlation of fault manifestations after negative correlation mapping; where x represents the independent variable and the time span is also the time interval.

[0064] In another embodiment of the present invention, the implementer may also combine the fitting deviation of the fitting line to express the correlation stability of the two abnormal node sequences and obtain the fault performance correlation degree; that is, the total distance from all sample points to the fitting line represents the fitting deviation. The smaller the total distance, the better the straight line fitting effect and the stronger the correlation stability. From an overall perspective, the more stable the correlation between the two data dimensions, the greater the correlation degree of the node pair and the greater the fault performance correlation degree.

[0065] Based on this, ,in It represents the total distance from all sample points to the fitting line. The total distance is small, and the negative correlation mapping is used to The greater the correction weight assigned, the greater the correlation of the fault manifestations.

[0066] Implementers can choose to increase the accuracy of the correlation between fault manifestations by adding a certain amount of calculation.

[0067] It should be noted that when the time span is zero, a very small positive value such as 0.01 is given to it to avoid the denominator being 0. The fitting straight line can use the least squares method, which is an existing technology and will not be described in detail.

[0068] Step S3: In the latest abnormal node sequence in the temperature dimension, according to the fault performance correlation of all node pairs corresponding to each time point, combined with the fluctuation characteristics of the mean value of the data abnormality coefficient of all node pairs, the abnormal factor of each time point is obtained; according to the change of the abnormal factor in the latest abnormal node sequence in the temperature dimension, combined with the temperature change in the meter box, the current target heat dissipation power is obtained and it is determined whether to issue an early warning.

[0069] Considering that the temperature dimension will match with multiple other dimensions, the correlation degree of all fault manifestations corresponding to the time points in the latest abnormal node sequence of the temperature dimension represents the overall abnormal correlation strength between temperature and other dimensions, that is, the possibility that all dimensions jointly show signs of faults; considering the fluctuation characteristics of the mean value of the data anomaly coefficient corresponding to all node pairs, it represents the stability and reliability of multi-dimensional abnormal signals. Therefore, according to the correlation degree of fault manifestations corresponding to all node pairs at each time point, combined with the fluctuation characteristics of the mean value of the data anomaly coefficient corresponding to all node pairs, the abnormal factors at each time point are obtained. By matching multi-dimensional data, the overall observability of the fault signal is enhanced, providing a basis for subsequent adjustment of heat dissipation power and early warning.

[0070] Preferably, in one embodiment of the present invention, the mean of the data anomaly coefficients of the node pair is used as the anomaly parameter of the corresponding node pair;

[0071] In the latest abnormal node sequence of the temperature dimension, the ratio of the mean of the fault performance correlation of all node pairs corresponding to each time point to the variance of the abnormal parameters of all node pairs is used as the abnormal factor of the corresponding time point.

[0072] Among them, the larger the mean of the fault performance correlation of all node pairs corresponding to the time point, the greater the overall intensity of the temperature dimension anomaly and the larger the anomaly factor; the larger the variance of the abnormal parameters of all node pairs corresponding to the time point, the worse the stability of the anomaly. The anomaly may be a short-term random fluctuation or environmental interference. The lower the credibility, the smaller the anomaly factor. Therefore, the variance of the abnormal parameter is in the denominator, and the mean of the fault performance correlation is in the numerator.

[0073] It should be noted that the variance is already an existing technology. When the variance is 0, it is given a very small positive parameter of 0.01 to prevent the denominator from being zero.

[0074] The abnormal factor is calculated based on the correlation between temperature and other data dimensions and the abnormal fluctuation characteristics, and characterizes the abnormal correlation between different dimensions. In the temperature dimension, the change of the abnormal factor reflects the abnormal accumulation trend of heat inside the meter box, showing the potential failure risk of the working state. At the same time, the temperature change directly reflects the real-time heat level change of the meter box. Therefore, according to the change of the abnormal factor in the latest abnormal node sequence in the temperature dimension, combined with the temperature change in the meter box, it can provide an accurate reference for the calculation of the heat dissipation power, reflect the actual thermal management needs of the equipment, obtain the current target heat dissipation power, and analyze the current abnormal risk of the meter box to determine whether to issue an early warning, avoid the one-sidedness of single-dimensional data, and avoid the misjudgment of the early warning signal due to inaccurate heat dissipation assessment of the meter box, improve the accuracy of the early warning, and provide long-distance indication for the work of the meter box in the power grid.

[0075] Preferably, in one embodiment of the present invention, considering that at both ends of the abnormal node sequence, the larger the temperature data of the tail segment compared with the starting segment, and the larger the abnormal factor, the more likely heat accumulation is inside the meter box, and the heat accumulation causes the working state of the internal components to gradually deviate from the normal range, the more it is necessary to increase the heat dissipation power, and the larger the target heat dissipation power;

[0076] Based on this, in the latest abnormal node sequence in the temperature dimension, the product of the abnormal factor with the largest time series and the corresponding temperature data value is used as the first numerator, the product of the abnormal factor with the smallest time series and the corresponding temperature data value is used as the first denominator, and the fractional ratio of the first numerator to the first denominator is used as the adjustment coefficient;

[0077] The product of the adjustment coefficient and the current heat dissipation power is taken as the current target heat dissipation power.

[0078] It should be noted that the target heat dissipation power is not greater than the maximum heat dissipation power of the meter box, that is, when the target heat dissipation power is greater than the maximum heat dissipation power that the meter box can implement, the meter box is adjusted to the maximum heat dissipation power to effectively deal with the heat accumulation problem inside the meter box and ensure the stability and safety of equipment operation.

[0079] Preferably, in one embodiment of the present invention, considering that the growth trend of temperature data and abnormal factors is strong, it means that heat accumulation may have significantly affected the performance of components inside the meter box, and the current state of the meter box may be close to the critical point requiring intervention. Further delayed processing may lead to irreversible failure. When the adjustment coefficient is greater than the preset heating threshold, a warning signal is issued.

[0080] As an example, the adjustment coefficient is linearly normalized and the fever threshold is preset to 0.8.

[0081] It should be noted that when the latest abnormal node sequence in the temperature dimension is an empty sequence, no dynamic time warping matching is performed, the heat dissipation power is not adjusted, and no warning signal is issued.

[0082] An embodiment of the present invention further provides an electric meter box, which includes a memory, a processor and a computer program, wherein the memory is used to store the corresponding computer program, and the processor is used to run the corresponding computer program. When the computer program runs in the processor, it can implement the automatic fault troubleshooting and detection method of the electric meter box described in steps S1-S3.

[0083] In summary, in order to solve the technical problem that the existing multi-dimensional data-based monitoring of electric meter boxes is prone to misjudgment, the present invention proposes an automatic fault troubleshooting detection method for electric meter boxes and an electric meter box. The present invention first obtains the multi-dimensional electrical time series data sequence of the electric meter box; further, according to the local fluctuation characteristics of the data, the data anomaly coefficient at each time point is obtained, and the abnormal node sequence is obtained; further, based on dynamic time warping, node pairs of different abnormal node sequences are obtained, and the abnormal factor is obtained according to the change characteristics of the mean value of the data anomaly coefficient of the node pair over the time span, combined with the fluctuation characteristics of the mean value of the data anomaly coefficient; finally, according to the change of the abnormal factor, combined with the temperature change in the electric meter box, the current target heat dissipation power is obtained and it is determined whether to issue an early warning.

[0084] It should be noted that the sequence of the above embodiments of the present invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0085] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for automatically troubleshooting and detecting faults of an electric meter box, characterized in that: The method comprises: Obtain a multi-dimensional electrical time series data sequence of an electric meter box; select the latest time series data sequence in any dimension as a target sequence; obtain a data anomaly coefficient at each time point according to the local fluctuation characteristics of the data at each time point in the target sequence; obtain an abnormal node sequence in the target sequence based on the data anomaly coefficient; Match the latest abnormal node sequences in the temperature dimension and any other dimension through dynamic time warping to obtain node pairs; obtain the fault performance correlation of each node pair according to the change characteristics of the mean value of the data abnormality coefficient of the node pair over the time span; In the abnormal node sequence with the latest temperature dimension, according to the fault performance correlation of all node pairs corresponding to each time point, combined with the fluctuation characteristics of the mean value of the data abnormal coefficient corresponding to all node pairs, the abnormal factor at each time point is obtained; according to the change of the abnormal factor in the abnormal node sequence with the latest temperature dimension, combined with the temperature change in the meter box, the current target heat dissipation power is obtained and it is determined whether to issue an early warning; The method for obtaining the fault manifestation correlation degree includes: The mean of the data anomaly coefficient of the node pair is used as the ordinate and the time span is used as the abscissa to form a two-dimensional sample space; the slope of the fitting straight line of all sample points in the two-dimensional sample space is obtained; According to the ratio of the mean value of the data anomaly coefficient of each node pair to the time span and the similarity characteristics of the slope, the fault manifestation correlation of each node pair is obtained; After negative correlation mapping between the ratio of the mean value of the data anomaly coefficient of each node pair to the time span and the absolute value of the difference between the slopes, the mapping value is used as the fault manifestation correlation degree of each node pair; The method for obtaining the abnormal factor includes: Taking the mean of the data anomaly coefficients of the node pair as the anomaly parameter of the corresponding node pair; In the abnormal node sequence of the latest temperature dimension, the ratio of the mean of the fault performance correlation of all node pairs corresponding to each time point to the variance of the abnormal parameters of all node pairs is used as the abnormal factor of the corresponding time point; The method for obtaining the target heat dissipation power includes: In the abnormal node sequence with the latest temperature dimension, the product of the abnormal factor with the largest time series and the corresponding temperature data value is used as the first numerator, the product of the abnormal factor with the smallest time series and the corresponding temperature data value is used as the first denominator, and the fractional ratio of the first numerator to the first denominator is used as the adjustment coefficient; The product of the adjustment coefficient and the current heat dissipation power is used as the current target heat dissipation power.

2. The method for automatically troubleshooting and detecting faults of an electric meter box according to claim 1, characterized in that: The method for obtaining the data anomaly coefficient includes: The derivative of each data on the data curve of the target sequence is obtained; according to the difference characteristics between the derivative of each data and the mean of all the derivatives, combined with the absolute value of the mean of all the derivatives, the data anomaly coefficient of each data corresponding to the time point is obtained; the difference characteristics between the derivative of each data and the mean of all the derivatives are positively correlated with the data anomaly coefficient; the absolute value of the mean of all the derivatives is negatively correlated with the data anomaly coefficient.

3. The method for automatically troubleshooting and detecting faults of an electric meter box according to claim 2, characterized in that: The method for obtaining the abnormal node sequence includes: The time point when the data anomaly coefficient is greater than a preset anomaly threshold is marked as an abnormal node; and all the abnormal nodes in the target sequence constitute an abnormal node sequence.

4. The method for automatically troubleshooting and detecting faults of an electric meter box according to claim 1, characterized in that: The method for determining whether to issue an early warning comprises: When the adjustment coefficient is greater than a preset fever threshold, an early warning signal is issued.

5. The method for automatically troubleshooting and detecting faults of an electric meter box according to claim 1, characterized in that: The dimensions of the time series data sequence include at least temperature, humidity, voltage and current; the data of each dimension becomes equal in length after spline interpolation processing.

6. An electric meter box, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the automatic fault troubleshooting and detection method for an electric meter box as described in any one of claims 1 to 5 are implemented.

Images