Intelligent electric meter fault arc detection method based on intelligent sensing algorithm

Abstract

The invention relates to the field of intelligent electric meters, in particular to an intelligent electric meter fault arc detection method based on an intelligent sensing algorithm, which comprises the following steps of: determining a circuit load state according to a dynamic characteristic value of a related load of an electric energy meter and a related load quantity; determining a sampling mode as prediction optimization sampling or sampling by adopting a reference sampling frequency according to a circuit load state; determining a first-order sampling frequency based on the number of loads to be analyzed during prediction and optimization of sampling; under the condition of primary optimization, performing one-class optimization analysis on each load to be analyzed, and determining whether to adjust the sampling frequency of the load to be analyzed or not according to the node completion degree average value; under the secondary optimization condition, determining a sampling setting strategy according to the interval balance degree; according to the invention, the sampling precision of the sampling mode is improved, so that the accuracy of fault arc detection is improved.

Description

Technical Field

[0001] The present invention relates to the field of smart meters, and in particular to a smart meter fault arc detection method based on an intelligent sensing algorithm. Background Art

[0002] Arc faults are a common electrical fault in power systems, posing a serious threat to their safe operation. Currently, arc fault detection technology based on intelligent sensing algorithms accurately identifies arc faults by collecting electrical signals such as current and voltage and combining them with signal processing, pattern recognition, and machine learning algorithms. However, in practical applications, the accuracy of arc fault detection in industrial scenarios is easily affected by environmental complexity, load diversity, and the characteristics of the arc signal itself, making arc fault detection difficult. Therefore, achieving accurate arc fault detection in industrial scenarios is a technical problem that urgently needs to be addressed by researchers in this field.

[0003] Chinese Patent Publication No. CN107370124B discloses an arc fault detection method, comprising: a data sampling step, in which data is sampled according to a predetermined sampling frequency and sampling period; a sampled data preprocessing step, in which the sampled data is preprocessed and stored; a sampled data processing step, in which the preprocessed sampled data is processed, and the fault type is pre-determined based on the processing result and a corresponding fault judgment subroutine is started, or a trip instruction is issued; an action execution step, in which a trip action is executed or not executed based on the processing result of the sampled data processing. If the trip action is executed, the entire arc fault detection method ends. If the trip action is not executed, the method returns to the data sampling step to continue data sampling and repeat the above process. It can be seen that the above technical solution has the following problems: it does not consider the influence of influencing factors in industrial scenarios on fault arc detection, and the predetermined sampling frequency and sampling period make it difficult for the sampled data obtained by a single sampling method to meet the actual detection efficiency requirements, thereby reducing the efficiency of fault arc detection. Summary of the Invention

[0004] To this end, the present invention provides a smart meter arc fault detection method based on an intelligent sensing algorithm to overcome the fact that the existing technology does not consider the impact of influencing factors in industrial scenarios on arc fault detection. The sampling data obtained by a single sampling method is difficult to meet the actual detection efficiency requirements, thereby reducing the arc fault detection efficiency.

[0005] To achieve the above objectives, the present invention provides a method for detecting arc faults in smart meters based on an intelligent sensing algorithm, comprising:

[0006] Determine the circuit load state according to the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads, and determine the sampling method as predictive optimization sampling or sampling using a reference sampling frequency according to the circuit load state;

[0007] When predicting optimization sampling, the initial sampling frequency is determined based on the number of loads to be analyzed;

[0008] Under the one-time optimization condition, a type of optimization analysis is performed for each load to be analyzed, and whether the sampling frequency of the load to be analyzed should be adjusted is determined based on the comparison result of the average node completion degree with the preset average node completion degree;

[0009] Under the condition of quadratic optimization, the sampling setting strategy is determined as uniform interval sampling or variable interval sampling according to the interval balance.

[0010] Furthermore, the circuit load state is determined according to the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads. The circuit load state includes:

[0011] A first circuit load state in which the dynamic characteristic value is greater than a preset dynamic characteristic value and the number of related loads is greater than a preset number of related loads;

[0012] a second circuit load state in which the dynamic characteristic value is less than or equal to a preset dynamic characteristic value or the number of related loads is less than or equal to a preset number of related loads;

[0013] Further, a sampling mode is determined according to the circuit load state;

[0014] When the circuit load state is in the first circuit load state, the sampling mode is predictive optimization sampling;

[0015] When the circuit load state is in the second circuit load state, the sampling method is to perform sampling using the reference sampling frequency.

[0016] Furthermore, when the sampling method is predictive optimization sampling, the primary sampling frequency is determined based on the number of loads to be analyzed;

[0017] The primary sampling frequency is positively correlated with the number of loads to be analyzed.

[0018] Furthermore, under one optimization condition, a type of optimization analysis is performed for each load to be analyzed to obtain the node completion degree of a type of related equipment corresponding to each load to be analyzed;

[0019] If the average node completion degree is greater than the preset node completion degree average, the primary sampling frequency is increased and the increased primary sampling frequency is recorded as the intermediate sampling frequency;

[0020] Among them, the primary optimization condition is that the frequency sampling method is predictive optimization sampling and the initial sampling frequency is determined and completed.

[0021] Furthermore, under the secondary optimization condition, a two-class optimization analysis is performed for each load to be analyzed to obtain several prediction intervals, and the sampling setting strategy is determined according to the interval balance;

[0022] If the interval balance is greater than the preset interval balance, the sampling setting strategy is equal interval sampling;

[0023] If the interval balance is less than or equal to the preset interval balance, the sampling strategy is set to variable interval sampling;

[0024] The secondary optimization condition is that the intermediate sampling frequency is confirmed to be complete or it is determined that there is no need to adjust the primary sampling frequency.

[0025] Furthermore, the second type of optimization analysis for a single load to be analyzed includes:

[0026] Detecting the route congestion coefficient and route length of each Class II related device of the load to be analyzed to determine the initial estimated time point of each Class II related device;

[0027] The preset interval corresponding to the initial estimated time point is determined based on the transportation conflict degree and range change amount of each Class II related equipment.

[0028] Furthermore, the interval balance degree is confirmed by extracting each preset interval corresponding to each load to be analyzed, extracting the initial estimated time point corresponding to each preset interval, and calculating the time interval between each adjacent initial estimated time point. The calculation formula of the interval balance degree S is:

[0029]

[0030] Where Si is the i-th time interval, i=1, 2, 3, ..., m, where m is the total number of time intervals.

[0031] Furthermore, when the sampling setting strategy is equal-interval sampling, data sampling is performed using a primary sampling frequency or a mid-order sampling frequency within a preset sampling period.

[0032] Furthermore, in the variable interval sampling, within the preset sampling period, the data is sampled at the first section using the primary sampling frequency or the intermediate sampling frequency, and the data is sampled at the second section using the final sampling frequency.

[0033] Compared with the prior art, the beneficial effect of the present invention lies in that, in the technical solution of the present invention, the circuit load state is determined according to the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads, and the dynamic characteristic value of the relevant load and the number of relevant loads are used to reflect the possible influence of the relevant load on the fault arc detection, and different sampling methods are determined accordingly, thereby avoiding the problem that the data obtained by the single sampling method in the prior art is affected by harmonic current and current distortion, which makes it difficult to support the fault arc detection requirements. The sampling method of the present invention can more effectively meet the load conditions of the circuit in the actual scenario, thereby improving the data sampling efficiency and data validity.

[0034] Furthermore, in the technical solution of the present invention, when a type of optimization analysis is performed on a single load to be analyzed under one optimization condition, when the average node completion degree of each type of related equipment is greater than the preset average node completion degree, the sampling frequency is increased and adjusted. The average node completion degree reflects the subsequent work busyness of the related load corresponding to each type of related equipment. Taking into account the impact of a type of related equipment on the operating status of the load to be analyzed, the primary sampling frequency can be adjusted predictively, so that the primary sampling frequency is more in line with the actual working scenario. Compared with the technical means of adjusting the sampling method when the detection effect is poor in the prior art, the present invention avoids the delay problem of the prior art, improves data sampling efficiency and data validity, and further improves the fault arc detection efficiency of the present invention.

[0035] Furthermore, in the technical solution of the present invention, under the condition of secondary optimization, two types of optimization analysis are performed for each load to be analyzed to obtain several prediction intervals, and different sampling setting strategies are determined based on the comparison results of the interval balance and the preset interval balance. The interval balance reflects whether there is an impact between the working states of the two types of related equipment, making the specific sampling settings more targeted, avoiding the problem of a single sampling method leading to a decrease in the accuracy of fault arc detection, and thereby improving the validity of the sampling data.

[0036] Furthermore, the technical solution of the present invention detects the route congestion coefficient and route length of each Class II related equipment of the load to be analyzed to determine the initial estimated time point of each Class II related equipment, and determines the preset interval according to the transportation conflict degree and range iteration amount of each Class II related equipment. The estimated working conditions of the Class II related equipment are reflected by the route congestion coefficient and route length, and the time for completing the task of the Class II related equipment is effectively analyzed and adjusted by the transportation conflict degree and range iteration amount, thereby effectively predicting the working conditions of subsequent related loads, taking into account the influence of the Class II related equipment on the operating status of the load to be analyzed, thereby improving the collection efficiency, and further improving the fault arc detection efficiency of the present invention.

[0037] Furthermore, in the technical solution of the present invention, the extraction effect evaluation value corresponding to each feature extraction strategy is determined according to the feature partition coefficient and the feature fluctuation coefficient, and the feature extraction method is determined according to the difference in the evaluation values to select the feature extraction strategy with the largest extraction effect evaluation value for feature extraction or combined feature selection. The feature partition coefficient reflects the difference between different extracted features. The higher the degree of difference, the higher the effectiveness of the corresponding extracted features. The feature fluctuation coefficient reflects whether the circuit is in a stable state, which is conducive to effective analysis of the extraction effect evaluation value corresponding to each feature extraction strategy, avoiding the inability of a single extraction method to fully extract features, thereby improving the adequacy of feature extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 Schematic diagram of a smart meter arc fault detection method based on an intelligent sensing algorithm according to the present invention;

[0039] Figure 2 This is a flow chart of the present invention for determining the circuit load state according to the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads;

[0040] Figure 3 This is a flow chart of the present invention for determining a sampling method according to a circuit load state;

[0041] Figure 4 This is a flow chart of the present invention for determining a sampling setting strategy based on interval balance. DETAILED DESCRIPTION

[0042] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0043] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0044] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0045] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0046] See also Figures 1 to 4 As shown, the present invention provides a smart meter fault arc detection method based on intelligent sensing algorithm, comprising:

[0047] Determine the circuit load state according to the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads, and determine the sampling method as predictive optimization sampling or sampling using a reference sampling frequency according to the circuit load state;

[0048] When predicting optimization sampling, the initial sampling frequency is determined based on the number of loads to be analyzed;

[0049] Under the one-time optimization condition, a type of optimization analysis is performed for each load to be analyzed, and whether the sampling frequency of the load to be analyzed should be adjusted is determined based on the comparison result of the average node completion degree with the preset average node completion degree;

[0050] Under the condition of quadratic optimization, the sampling setting strategy is determined as uniform interval sampling or variable interval sampling according to the interval balance.

[0051] In the present invention, the electric energy meter is applied to industrial scenarios. The industrial scenarios include but are not limited to related loads, Class I related equipment, and Class II related equipment. The related loads are load equipment that will generate load harmonics when detecting fault arcs on the electric energy meter, for example, arc welding machines and equipment with inverters. Class I related equipment is production equipment that has a corresponding working relationship with the related loads and is fixed in position. Class II related equipment is movable transportation equipment that has a corresponding transportation relationship with the related loads. It can be understood that the working relationship is that the workpiece produced by the Class I related equipment needs to be transported to the related load for processing, and the transportation relationship is that the Class II related equipment needs to transport the workpiece to the related load. For example, if the related load is an arc welding machine for steel plate processing, then its corresponding Class I related equipment is a steel plate preparation system, the corresponding Class II related equipment is an AGV transport vehicle, and the workpiece is a steel plate. This is content that is easy to understand for those skilled in the art and will not be elaborated here. It can be understood that in the present invention, each related load and the Class I related equipment and Class II related equipment corresponding to the related load are set in advance by the user, and the brand and specifications of the electric energy meter are not limited here.

[0052] The present invention applies several historical records, and a single historical record includes a record of fault arc detection in at least one history and the corresponding detection value. The detection value includes but is not limited to the number of starts and stops, load running time, dynamic characteristic value, number of related loads, average node completion degree, interval balance, loading ratio, transportation conflict degree and range change amount, and there is a qualified mark corresponding to the historical record. The qualified mark records whether the fault arc detection effect corresponding to the historical record meets the user's needs. Among them, whether the user's needs are met is determined by the user based on the arc detection efficiency or arc detection accuracy. This is content that technical personnel in this field have already mastered and there is no need to elaborate here.

[0053] Specifically, the circuit load state is determined based on the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads. The circuit load state includes:

[0054] A first circuit load state in which the dynamic characteristic value is greater than a preset dynamic characteristic value and the number of related loads is greater than a preset number of related loads;

[0055] a second circuit load state in which the dynamic characteristic value is less than or equal to a preset dynamic characteristic value or the number of related loads is less than or equal to a preset number of related loads;

[0056] The present invention applies a cyclically repeated monitoring cycle, and the duration of the monitoring cycle is set by the user. It can be understood that the greater the user's demand for detection accuracy, the shorter the duration of the monitoring cycle. In the specific implementation of the present invention, the duration of a single monitoring cycle is 5 minutes. The dynamic characteristic value is confirmed in the following way: a dynamic characteristic value is determined at the end of each monitoring cycle, and the sum of the sub-dynamic values of each related load in the monitoring cycle is recorded as the dynamic characteristic value. For a single related load, the corresponding sub-dynamic value = (number of starts and stops / preset number of starts and stops) × α1 + (load running time / preset load running time) × α2, the number of starts and stops is the number of starts and stops of the related load in the most recent monitoring cycle, and the load running time is the working time of the related load in the most recent monitoring cycle, wherein α1 is the first dynamic weight coefficient, and α2 is the second dynamic weight coefficient. The values of α1 and α2 can be obtained by the user by means of, but not limited to, historical records, human experience settings or deep learning. It can be understood that the user can It is possible to learn historical records that meet user needs through a deep learning network, obtain the values of α1 and α2, and optimize them. No further details are given here. In the specific implementation of the present invention, α1=0.4, α2=0.6, and the values of the preset start and stop times and the preset load running time can be set by the user according to the actual application scenario. It can be understood that the start and stop times and the load running time can effectively reflect the influence of the relevant load on the fault arc detection accuracy. The greater the user's demand for fault arc detection accuracy, the smaller the values of the preset start and stop times and the preset load running time. A method for determining the preset start and stop times and the preset load running time is provided, and the corresponding start and stop times and load running time in the historical records that meet user needs are extracted. The abnormal values therein are screened out by a machine learning method, and the average values of the start and stop times and the load running time after removing the abnormal values are respectively recorded as the preset start and stop times and the preset load running time. In the specific implementation of the present invention, the preset start and stop times are 3 times, and the preset load running time is 5 minutes.

[0057] The preset dynamic characteristic value can be set by the user according to the actual application scenario. In the present invention, the dynamic characteristic value can effectively reflect the influence of the relevant load on the fault arc detection. The greater the user's demand for the accuracy of the fault arc detection, the smaller the preset dynamic characteristic value. A method for obtaining the preset dynamic characteristic value is provided to extract the dynamic characteristic value from the historical records that meet the user's needs, screen out the abnormal values therein through machine learning methods, and record the average value of the dynamic characteristic value after removing the abnormal values as the preset dynamic characteristic value.

[0058] The value of the preset number of related loads can be set by the user according to the actual application scenario. It is understandable that the current waveform of the nonlinear load during operation is easily superimposed on the current characteristics generated by the fault arc, resulting in unclear characteristics of the fault arc, increasing the complexity of detection, and thus increasing the difficulty of detection. Therefore, the higher the user's expectation for the accuracy of fault arc detection, the smaller the preset number of related loads. In the specific implementation of the present invention, the preset number of related loads is 30%.

[0059] Specifically, the sampling method is determined according to the circuit load state;

[0060] When the circuit load state is in the first circuit load state, the sampling mode is predictive optimization sampling;

[0061] When the circuit load state is in the second circuit load state, the sampling method is to perform sampling using the reference sampling frequency.

[0062] In the present invention, the larger the value of the reference sampling frequency is, the greater the sampling data accuracy is and the greater the data processing capability required is. The user can set it according to the actual application scenario and provide a reference sampling frequency value, reference sampling frequency = 500kHz.

[0063] Specifically, when the sampling method is predictive optimization sampling, the primary sampling frequency is determined based on the number of loads to be analyzed;

[0064] The primary sampling frequency is positively correlated with the number of loads to be analyzed.

[0065] The relevant load corresponding to at least one of the first-category related equipment and the second-category related equipment is recorded as the load to be analyzed.

[0066] Specifically, under one optimization condition, a type of optimization analysis is performed for each load to be analyzed to obtain the node completion degree of a type of related equipment corresponding to each load to be analyzed;

[0067] If the average node completion degree is greater than the preset node completion degree average, the primary sampling frequency is increased and the increased primary sampling frequency is recorded as the intermediate sampling frequency;

[0068] If the average node completion degree is less than or equal to the preset average node completion degree, it is determined that there is no need to adjust the primary sampling frequency;

[0069] Among them, the primary optimization condition is that the frequency sampling method is predictive optimization sampling and the initial sampling frequency is determined and completed.

[0070] For a single Class A related device, its corresponding node completion degree = the completed workload corresponding to the most recent production cycle. The average value of the node completion degrees corresponding to each Class A related device is recorded as the node completion average value. The completed workload is the number of workpieces produced by each Class A related device in the most recent production cycle. The present invention applies a value of the production cycle with a cyclic setting, and the user can set it according to the actual application scenario. The greater the production efficiency of a Class A related device, the smaller the value of the production cycle. One value is provided, and the length of the production cycle is 30 minutes.

[0071] The preset node completion average value can be adaptively set by the user according to the actual situation. It can be understood that the preset node completion average value can effectively reflect the impact of a type of related equipment on the operating status of the load to be analyzed. The comparison result of the node completion average value and the preset node completion average value makes it possible to adjust the primary sampling frequency numerically in advance, thereby improving the validity of the sampling data. A method for obtaining the preset node completion average value is provided, which extracts the node completion average value from the historical records that meet the user's needs, detects the outliers therein by the outlier method, and records the median value of all the node completion average values after removing the outliers as the preset node completion average value.

[0072] When the average node completion degree is greater than the preset average node completion degree, the difference between the average node completion degree and the preset average node completion degree is recorded as V, where V = average node completion degree - preset average node completion degree. When the primary sampling frequency is increased and adjusted, V is positively correlated with the increase in the primary sampling frequency. A specific increase method is provided, the intermediate sampling frequency = primary sampling frequency + k1×V, the increase = k1×V, k1 is the correlation coefficient, and the user can obtain it through historical records. How to obtain training samples and learn weight coefficients through historical records is content that is easy for technical personnel in this field to understand and will not be elaborated here. The present invention is provided with a maximum increase, which is 10kHz. When the calculated value of the increase is greater than 10kHz, the increase is taken according to the maximum increase.

[0073] Specifically, under the quadratic optimization condition, a two-class optimization analysis is performed for each load to be analyzed to obtain several prediction intervals, and the sampling setting strategy is determined based on the interval balance;

[0074] If the interval balance is greater than the preset interval balance, the sampling setting strategy is equal interval sampling;

[0075] If the interval balance is less than or equal to the preset interval balance, the sampling setting strategy is variable interval sampling; the secondary optimization condition is that the intermediate sampling frequency is confirmed to be completed or it is determined that there is no need to adjust the primary sampling frequency.

[0076] Specifically, the second type of optimization analysis for a single load to be analyzed includes:

[0077] Detect the route congestion coefficient and route length of each Class II related device of the load to be analyzed to determine the initial estimated time point of each Class II related device;

[0078] The preset interval is determined based on the transportation conflict degree and range change amount of each Class II related equipment.

[0079] For a single Class II related device of a load to be analyzed, there are several unfinished task points corresponding to the Class II related device. It can be understood that there may be multiple related loads with transportation relationships for the Class II related device. When the Class II related device receives a task instruction, it obtains the task point corresponding to the task instruction. The task point is the pickup point and the delivery point. The unfinished task point is the task point corresponding to the currently unfinished task instruction of the Class II related device, and a work route is generated based on the unfinished task point. The distance from the starting point to the load to be analyzed in the work route is recorded as the route length. Among them, how to generate a work route based on the unfinished task point is content that technicians in this field have already mastered, and no specific limitation is made here. A method for generating a work route based on the unfinished task point is provided. The working route is determined in the order in which the task instructions are received. If the current task instructions include task instruction 1, task instruction 2, and task instruction 3, the task instruction corresponding to the load to be analyzed is task instruction 3. The working route starts from the current location of the second-class related equipment and is in the order of the pick-up point of task instruction 1, the delivery point of task instruction 1, the pick-up point of task instruction 2, the delivery point of task instruction 2, the pick-up point of task instruction 3, and the delivery point of task instruction 3. The traversable routes between each adjacent point (including the starting point, the pick-up point, and the delivery point) are obtained, and the traversable route with the shortest path is recorded as a sub-route. The sum of the sub-routes is the working route. The length of the working route is the route length of the second-class related equipment.

[0080] The method for confirming the route congestion coefficient of a single Class II related device is to detect the number of Class II related devices on its working route and record it as the route congestion coefficient;

[0081] The initial estimated time point is the time point after the current moment with an estimated interval. Estimated time = (route length / estimated speed) + congestion value, congestion value = route congestion coefficient × L, where the value of the estimated speed can be set by the user. The user can count the time required to complete different work routes in the historical records to calculate the moving speed, and record the average moving speed as the estimated speed. A value of the estimated speed is provided, which is 80% of the maximum operating speed of the second-class related equipment in the absence of interference. L is the congestion base value, which is a constant. It can be understood that the larger the value of L, the greater the impact of the number of second-class related equipment on the working route on the working time of the second-class related equipment. A value of L is provided, L = 0.5, and the units of the estimated time and congestion value are both seconds.

[0082] Each initial estimated time point corresponds to a preset interval, which is a time segment with the initial estimated time point as the center point, and the unit of the preset interval is s; for a single Class II related equipment, it is recorded as the target related equipment, and its transportation conflict degree is the number of other movable equipment on the working route of the target related equipment whose transportation direction is opposite to that of the target related equipment. The movable equipment includes but is not limited to other Class II related equipment, manual transport carts and robots, etc. The range change amount is all Class II related equipment corresponding to other related loads whose loading ratio is less than the preset loading ratio within the preset range of the working route of the target related equipment. For a single Class II related equipment, its corresponding loading ratio is the weight of the workpiece it is currently loading / its maximum loading weight. The loaded weight can be obtained through a weight sensor. The preset loading ratio is a reference value for determining the current loading capacity of the Class II related equipment. It can be understood that the stronger the loading capacity of the Class II related equipment, the greater the probability of it receiving task instructions and the greater the impact on line transportation. The value of the preset loading ratio in the present invention is 50%.

[0083] The preset interval is positively correlated with the transport conflict degree and the range iteration amount. In the specific implementation of the present invention, the preset interval = initial interval × effective coefficient, the effective coefficient is the larger value of ratio A and ratio B, ratio A is the transport conflict degree / preset transport conflict degree, and ratio B is the range iteration amount / preset range iteration amount. The values of the initial interval, preset transport conflict degree and preset range iteration amount can be set by the user according to the actual scenario. It can be understood that the greater the user's demand for the accuracy of the preset interval, the smaller the values of the preset transport conflict degree and the preset range iteration amount, and the smaller the initial interval. The values of the initial interval, preset transport conflict degree and preset range iteration amount are provided. The initial interval is 60s, the preset transport conflict degree is 10, and the value of the preset range iteration amount is 8.

[0084] The interval balance degree is confirmed by extracting each preset interval corresponding to the load to be analyzed, extracting the initial estimated time point corresponding to each preset interval, and calculating the time interval between each adjacent initial estimated time point. The calculation formula of the interval balance degree S is:

[0085]

[0086] Among them, Si is the i-th time interval, and the order of i can be random and does not affect the detection results. i = 1, 2, 3, ..., m, where m is the total number of time intervals. The user can adaptively set the value of the preset interval balance degree based on actual conditions. It is understood that the interval balance degree can effectively reflect the impact of the operating status of the second-class related equipment on the accuracy of fault arc detection. The greater the user's demand for fault arc detection accuracy, the larger the value of the preset interval balance degree. A method for determining the value of the preset interval balance degree is provided. The interval balance degree is extracted from historical records that meet the user's requirements. Outliers are identified using the Z-Score method. The average value of the interval balance degree after removing the outliers is recorded as the preset interval balance degree.

[0087] Specifically, when the sampling setting strategy is equal-interval sampling, the primary sampling frequency or the intermediate sampling frequency is used for data sampling within the preset sampling period. If the average node completion value is greater than the preset node completion value, the intermediate sampling frequency is used. If it is determined that the primary sampling frequency does not need to be adjusted, the primary sampling frequency is used for data sampling.

[0088] Specifically, in the variable interval sampling, within the preset sampling period, the data is sampled at the primary sampling frequency or the intermediate sampling frequency for the first section, and the data is sampled at the final sampling frequency for the second section.

[0089] The preset sampling period is from the current moment to the next sampling setting strategy confirmation completion.

[0090] In the present invention, the specific data sampling method is set by the user. For example, if arc fault detection is achieved through current signal analysis, the sampled data is the current waveform. After sampling, the presence of an arc fault is determined based on abnormal characteristics in the current waveform (such as high-frequency oscillation and extended zero-break). Sampling is performed using the corresponding sampling frequency, and the specific waveform sampling length is set by the user. The greater the user's demand for sampled data validity, the longer the waveform sampling length. This is a matter that is readily understood by those skilled in the art.

[0091] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A method for detecting arc faults in smart meters based on intelligent sensing algorithms, characterized in that: include: Determine the circuit load state according to the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads, and determine the sampling method as predictive optimization sampling or sampling using a reference sampling frequency according to the circuit load state; When predicting optimization sampling, the initial sampling frequency is determined based on the number of loads to be analyzed; Under the one-time optimization condition, a type of optimization analysis is performed for each load to be analyzed, and whether the sampling frequency of the load to be analyzed should be adjusted is determined based on the comparison result of the average node completion degree with the preset average node completion degree; Under the condition of quadratic optimization, the sampling setting strategy is determined as uniform interval sampling or variable interval sampling according to the interval balance.

2. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 1, characterized in that: The circuit load status is determined based on the dynamic characteristic value of the relevant load of the electric energy meter and the number of relevant loads. The circuit load status includes: A first circuit load state in which the dynamic characteristic value is greater than a preset dynamic characteristic value and the number of related loads is greater than a preset number of related loads; a second circuit load state in which the dynamic characteristic value is less than or equal to a preset dynamic characteristic value or the number of related loads is less than or equal to a preset number of related loads; 3. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 2, characterized in that: Determine the sampling method according to the circuit load status; When the circuit load state is in the first circuit load state, the sampling mode is predictive optimization sampling; When the circuit load state is in the second circuit load state, the sampling method is to perform sampling using the reference sampling frequency.

4. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 3, characterized in that: When the sampling method is predictive optimization sampling, the initial sampling frequency is determined based on the number of loads to be analyzed; The primary sampling frequency is positively correlated with the number of loads to be analyzed.

5. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 4, characterized in that: Under the one-time optimization condition, a type of optimization analysis is performed for each load to be analyzed to obtain the node completion degree of a type of related equipment corresponding to each load to be analyzed; If the average node completion degree is greater than the preset node completion degree average, the primary sampling frequency is increased and the increased primary sampling frequency is recorded as the intermediate sampling frequency; Among them, the primary optimization condition is that the frequency sampling method is predictive optimization sampling and the initial sampling frequency is determined and completed.

6. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 5, characterized in that: Under the quadratic optimization condition, a two-class optimization analysis is performed for each load to be analyzed to obtain several prediction intervals, and the sampling setting strategy is determined based on the interval balance; If the interval balance is greater than the preset interval balance, the sampling setting strategy is equal interval sampling; If the interval balance is less than or equal to the preset interval balance, the sampling strategy is set to variable interval sampling; The secondary optimization condition is that the intermediate sampling frequency is confirmed to be complete or it is determined that there is no need to adjust the primary sampling frequency.

7. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 6, characterized in that: The second type of optimization analysis for a single load to be analyzed includes: Detecting the route congestion coefficient and route length of each Class II related device of the load to be analyzed to determine the initial estimated time point of each Class II related device; The preset interval corresponding to the initial estimated time point is determined based on the transportation conflict degree and range change amount of each Class II related equipment.

8. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 7, characterized in that: The interval balance degree is confirmed by extracting each preset interval corresponding to each load to be analyzed, extracting the initial estimated time point corresponding to each preset interval, and calculating the time interval between each adjacent initial estimated time point. The calculation formula of the interval balance degree S is: Where Si is the i-th time interval, m is the total number of time intervals.

9. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 8, characterized in that: When the sampling setting strategy is equal interval sampling, data sampling is performed using the primary sampling frequency or the intermediate sampling frequency within the preset sampling period.

10. The method for detecting arc faults in smart meters based on intelligent sensing algorithms according to claim 8, characterized in that: In variable interval sampling, within a preset sampling period, data sampling is performed at the primary sampling frequency or the intermediate sampling frequency for the first section, and data sampling is performed at the final sampling frequency for the second section.

Images