Semi-solid state circuit breaker with full electric quantity data intelligent acquisition function

Abstract

The invention relates to the technical field of electric power data analysis, in particular to a semi-solid state circuit breaker with a full electric quantity data intelligent acquisition function. The semi-solid state circuit breaker comprises a data acquisition unit which is used for respectively acquiring different types of full electric quantity data in a preset time period; the noise possibility analysis unit is used for acquiring same-position data corresponding to each piece of data and calculating the noise possibility of each piece of data; the data initial segmentation unit is used for obtaining the turning moment possibility of each piece of data, and then segmenting the full electric quantity data to obtain initial data segments; the data segmentation correction unit is used for calculating the periodic rule similarity between the two initial data segments and then merging the initial data segments to obtain a final data segment; and the missing data complementing unit is used for setting a sliding window, acquiring the influence weight of each piece of data in other final data segments by utilizing the sliding window, and complementing the missing data. The missing data in the full electric quantity data can be accurately complemented.

Description

Technical Field

[0001] The present invention relates to the technical field of power data analysis, and particularly to a semi-solid circuit breaker with an intelligent full-power data acquisition function. Background Art

[0002] A semi-solid circuit breaker is a new type of power protection device that combines traditional mechanical switches and electronic switches. It not only has the high breaking capacity of traditional circuit breakers but also includes the high-speed response characteristics of solid-state devices. Full-power data acquisition is to perform high-precision, synchronous, and continuous data acquisition on multi-dimensional data such as current, voltage, power, harmonics, and temperature during the operation of the circuit breaker. Using these acquired data can achieve rapid fault identification and response, prediction and maintenance of circuit anomalies, and analysis and optimization of power quality, etc., improving power supply quality and power consumption safety.

[0003] Functions such as circuit anomaly prediction and fault response of semi-solid circuit breakers all rely on the analysis of the acquired data. When situations such as poor contact of connectors, electromagnetic interference, and communication anomalies occur during the acquisition process, it often causes discontinuous data acquisition and data loss problems. Due to the different operating states of the circuit, the data change laws are different. Traditional data interpolation methods cannot combine the data change laws of different states, and the obtained interpolation results may deviate greatly from the true values, thus affecting the accuracy of circuit anomaly prediction. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a semi-solid circuit breaker with an intelligent full-power data acquisition function, and the specific technical solution adopted is as follows: An embodiment of the present invention provides a semi-solid circuit breaker with an intelligent full-power data acquisition function, and the semi-solid circuit breaker includes: A data acquisition unit, configured to respectively acquire different types of full-power data within a preset time period; A noise possibility analysis unit, configured to obtain the same-position data corresponding to each data from the data within a window centered on each data; calculate the noise possibility of each data according to the same-position data corresponding to one data; A data initial segmentation unit, configured to obtain the turning moment possibility of a data based on the slope values of the two adjacent data corresponding to the data and the noise possibility of the data; segment the full-power data according to the turning moment possibilities of the data at the same moment in different types of full-power data to obtain initial data segments; A data segmentation correction unit, configured to calculate the periodic law similarity according to the time length difference and data difference between two initial data segments; merge the initial data segments according to the periodic law similarity between every two initial data segments to obtain final data segments; The missing data completion unit is used to set a sliding window based on the length of the final data segment where a missing data is located within a full power data, and slide the sliding window on other final data segments to obtain the influence weights of each data within the other final data segments; and complete the missing data based on each data within the other final data segments and its influence weights.

[0005] Preferably, obtaining the data at the same position corresponding to each data within the window centered on each data includes: In a full power data, obtain the sum of the absolute values of the differences between the data other than the data itself within the window corresponding to a data and the data at the corresponding position within the window of another data, then add it to a hyperparameter and take the reciprocal to obtain a reciprocal result, and normalize the reciprocal result to obtain the possibility that the other data is the data at the same position in different periods of this data; if the possibility that the other data is the data at the same position in different periods of this data is greater than the first reference threshold, then the other data is the data at the corresponding position of this data.

[0006] Preferably, calculating the noise possibility of a data based on the data at the same position corresponding to this data includes: Obtain the average value of the absolute values of the differences between a data and its corresponding data at the same position and normalize it to obtain the noise possibility of this data.

[0007] Preferably, obtaining the turning moment possibility of a data based on the slope values corresponding to two adjacent data of this data and the noise possibility of this data includes: Within a full power data, obtain the time domain graph corresponding to this full power data, and obtain the absolute value of the difference between the slopes corresponding to two adjacent data before and after a data in the time domain graph and normalize it to obtain the slope difference; multiply the slope difference by the difference between a preset value and the noise possibility of this data to obtain the turning moment possibility of this data.

[0008] Preferably, segmenting the full power data to obtain initial data segments according to the turning moment possibility of the data at the same moment in different types of full power data includes: Calculate the mean value of the turning moment possibilities of the data at the same moment in different types of full power data, denoted as the average turning possibility at this moment; if the average turning possibility at a moment is greater than the second reference threshold, then this moment is a suspected working state change moment, and use the suspected working state change moment to segment the full power data to obtain initial data segments.

[0009] Preferably, calculating the periodic law similarity according to the time length difference and data difference between two initial data segments includes: The absolute value of the difference in time length between two initial data segments is added to a hyperparameter and then the reciprocal is taken to obtain the time length similarity eigenvalue; the absolute value of the difference in a full power data at a certain moment between two initial data segments is obtained, which is denoted as the difference in this type of full power data at this moment. The sum of the absolute values of the differences in this type of full power data between each two adjacent moments is added to a hyperparameter and then the reciprocal is taken to obtain the data similarity eigenvalue between the two initial data segments corresponding to this type of full power data. The sum of the data similarity eigenvalues between the two initial data segments corresponding to each type of full power data is obtained to get the comprehensive data similarity eigenvalue between the two initial data segments; the time length similarity eigenvalue and the comprehensive data similarity eigenvalue are multiplied and normalized to obtain the periodic pattern similarity between the two initial data segments.

[0010] Preferably, according to the periodic pattern similarity between each two initial data segments, the initial data segments are merged to obtain the final data segments, including: Starting from the first initial data segment, obtain the initial data segments whose periodic pattern similarity with the first initial data segment is greater than the third reference threshold, which are denoted as suspected similar data segments; arrange the suspected similar data segments in chronological order to obtain the first sequence, and obtain the time interval between the starting moments of two adjacent suspected similar data segments in the first sequence as the adjacent interval between the two adjacent suspected similar data segments; if the adjacent interval between each two adjacent suspected similar data segments in the first sequence is less than or equal to the time interval threshold, then all the suspected similar data segments in the first sequence are the periodic data segments corresponding to the first initial data segment; if there are adjacent suspected similar data segments with an adjacent interval greater than the time interval threshold, denote the adjacent interval between such adjacent suspected similar data segments as the discontinuous interval, locate the first discontinuous interval in chronological order, and at this time, the suspected similar data segments before the first discontinuous interval are the periodic data segments corresponding to the first initial data segment; and so on, starting from the first initial data segment among the remaining initial data segments in chronological order, find the periodic data segments corresponding to the first initial data segment among the remaining initial data segments until all the periodic data segments are obtained; merge the two adjacent periodic data segments that are periodic data segments in the initial data segments in chronological order to obtain the final data segments.

[0011] Preferably, a sliding window is used to slide on other final data segments to obtain the influence weights of each data in the other final data segments, including: Slide a sliding window over the final data segment to which a piece of data belongs, with a sliding step size being a preset value, obtain the average of the absolute values of the differences between each piece of data within the sliding window and the corresponding data in the final data segment to which a missing piece of data belongs, and take the reciprocal to obtain the coincidence degree corresponding to this sliding window; obtain the maximum value among the coincidence degrees corresponding to each sliding window as the maximum coincidence degree; multiply the maximum coincidence degree by the difference between the preset value and the noise possibility of this data to obtain the influence weight of this data.

[0012] Preferably, filling in the missing data based on each piece of data and its influence weight in other final data segments includes: Obtain the influence weight of each piece of data in other final data segments except for the data in the final data segment where a missing piece of data is located, multiply the influence weight of each piece of data in other final data segments by each piece of data in other final data segments respectively to obtain the updated data corresponding to each piece of data in other final data segments; fill in the missing data according to the updated data corresponding to each piece of data in other final data segments and the LSTM algorithm.

[0013] The embodiments of the present invention have at least the following beneficial effects: This application collects different types of full power data, then analyzes the data at the same positions of each piece of data of different types of full power data, and then obtains the noise possibility of each piece of data, which is used to exclude the influence of noise when interpolating missing data, thereby improving the accuracy of filling in missing data; further, obtain the turning moment possibility of this data through the slope values corresponding to two adjacent pieces of data of each piece of data and the noise possibility of this data, and then segment the full power data to obtain initial data segments, and then calculate the periodic law similarity between every two initial data segments to merge the initial data segments to obtain final data segments. Taking into account both data changes and periodic changes to segment the data can effectively avoid misidentifying data changes caused by accidental factors as the change law of this segment of data, making the final data segments obtained after segmentation more accurate. Finally, by setting a sliding window with the same length as the final data segment where the missing data is located, it is used to identify the data in other final data segments with a similar change law to the missing data, and adjust the influence weight of the data in other final data segments on the interpolation result of the missing data accordingly, which can effectively improve the accuracy of data interpolation and is beneficial to improving the effectiveness of anomaly identification and warning. Description of the Drawings

[0014] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0015] Figure 1 This is a unit block diagram of a semi - solid circuit breaker with an intelligent full - power data acquisition function provided by an embodiment of the present invention. Detailed implementation manners

[0016] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following combines the drawings and preferred embodiments to detail the specific implementation manners, structures, features and effects of a semi - solid circuit breaker with an intelligent full - power data acquisition function proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0018] The following specifically describes the specific solution of a semi - solid circuit breaker with an intelligent full - power data acquisition function provided by the present invention with reference to the drawings.

[0019] Embodiment: The main application scenario of the present invention is as follows: During the data acquisition process of the semi - solid circuit breaker, problems such as missing acquired data may occur due to influencing factors such as poor contact and transmission interference, which will further affect the recognition and early warning of abnormal situations in the current monitored circuit and the safety of the circuit operation.

[0020] Please refer to Figure 1 , which shows a unit block diagram of a semi - solid circuit breaker with an intelligent full - power data acquisition function provided by an embodiment of the present invention. The semi - solid circuit breaker includes the following units: A data acquisition unit for respectively acquiring different types of full - power data within a preset time period.

[0021] The full - power data includes various types of data. In an embodiment of the present application, five types of full - power data are acquired, including current data, voltage data, power data, harmonic data, and temperature data. Therefore, a current transformer and a resistor divider need to be set in the semi - solid circuit breaker to acquire current data and voltage data respectively. The power data is obtained by calculating the product of current and voltage in the controller. The harmonic data is obtained using an FFT module. Since some abnormal manifestations of the circuit will also be significantly reflected in the temperature, a temperature sensor also needs to be set to monitor the temperature data. In order for the circuit breaker to more quickly identify and respond to circuit abnormal situations, the monitoring frequencies of current, voltage, power, and harmonic data are 10 kHz, while the temperature changes relatively slowly, so the temperature data is monitored once every 1 s.

[0022] Further, for the convenience of analysis, it is necessary for the monitoring moments of different types of data to correspond one by one. Since the monitoring frequency of temperature data is relatively low, the temperature data at one monitoring moment is used to represent the temperature value within 1 s after the monitoring of the regular customer, so as to achieve the purpose of data filling and make the full power data of each type align. Among them, since the value ranges of the full power data of different types are different, for the convenience of calculation and comparison, the data in each type of full power data obtained is normalized, and subsequent calculations are performed on the normalized data values. The time for collecting the full power data is a preset period, and the preset period is one day. In this application, the different types of full power data analyzed are the data collected in the most recent day.

[0023] The noise possibility analysis unit is used to obtain the co-located data corresponding to each data from the data within the window centered on each data; calculate the noise possibility of each data according to the co-located data corresponding to one data.

[0024] Due to interference factors such as electromagnetic interference, it is possible that noise data with data mutations may appear in the collected full power data of different types. These noise data will significantly deviate from the normal fluctuation signal and differ greatly from the real data, and have no reference significance. Therefore, the possibility that the historical data is noise data can be calculated first according to the degree of conformity between the historical data and the data change law within its time period. Since the data change characteristics of different working states of the circuit are different, the historical data in the same working state as the missing data is more in line with the data characteristics of the missing data position. And the closer the historical data is to the missing data in time, the closer the circuit state in its time period is. Therefore, according to the possibility that the historical data is in a similar working state as the current missing data, combined with the possibility that the historical data is noise data, the influence degree of the historical data on the interpolation result of the current missing data is calculated.

[0025] First, analyze the data in the full power data of different types to obtain the possibility that each data is noise data. Under normal conditions, the change range of the monitored data is small and changes regularly with a certain trend. While the noise data often shows sharp mutation signals. Therefore, according to the degree of conformity between each data and the data change characteristics within its time period, the possibility that each historical data is noise data is calculated.

[0026] To facilitate the analysis of the data change characteristics at the corresponding moment of each data, a window is established centered on each data according to the data monitoring frequency and the frequency of data change. Preferably, the time length of the window in the present invention is 1 s, and the implementer can adjust the time length of the window according to the actual situation to analyze the characteristics of the data located at the center of the window based on the data within the window.

[0027] According to the physical characteristics of the power grid monitored by the circuit breaker, the monitored data usually has obvious periodicity. Therefore, when not affected by noise, the values of the same-position data in different cycles of a full-power data should be close. Thus, according to the proximity of each data to the same-position data of the data to be analyzed in the corresponding window, the data in the same position in different cycles from the data to be analyzed is identified.

[0028] Next, the data in the window centered on each data is used to obtain the same-position data corresponding to each data. Specifically, in a full-power data, the sum of the absolute values of the differences between the data other than the data itself in the window corresponding to one data and the data at the corresponding positions in the window of another data is obtained, and then added to the hyperparameter and inverted to obtain an inverted result. The inverted result is normalized to obtain the possibility that another data is the same-position data in a different cycle of this data.

[0029] Furthermore, a first reference threshold is set, where the value of the first reference threshold is 0.8. The implementer can adjust the first reference threshold according to data statistics from multiple experiments. If, in a full-power data, the possibility that another data is the same-position data in a different cycle of a data is greater than the first reference threshold, then another data is the corresponding same-position data of this data. Thus, the same-position data corresponding to each data in the same full-power data can be obtained.

[0030] The calculation model for the possibility that another data is the same-position data in a different cycle of a data is: , where, represents the possibility that the j-th data in the c-th full-power data is the same-position data in a different cycle of the i-th data, norm is the normalization function, represents the number of data included in the window corresponding to the i-th data (excluding the i-th data itself); represents the value of the o-th data in the window corresponding to the i-th data in the c-th full-power data (the o-th data cannot be the i-th data itself), represents the value of the o-th data in the window corresponding to the j-th data in the c-th data, represents the absolute value of the difference between the data at the same position in the window corresponding to the i-th data and the window corresponding to the j-th data, It shows that the smaller the sum of the absolute values of the differences, the closer the data changes in the windows where these two data are located, that is, the more likely they are to be in the same position in different cycles. ε is a hyperparameter used to ensure that the denominator of the fraction is not zero, and here it is set to 0.01.

[0031] If a piece of data is not noise data, then the value of this data should be close to the value of the corresponding co-located data obtained according to the above process. Otherwise, it indicates that the value of this data significantly deviates from the variation law of the current data, that is, it may be noise data. Thus, the noise possibility of a piece of data is calculated based on the co-located data corresponding to this data.

[0032] Specifically, the average value of the absolute values of the differences between a piece of data and its corresponding co-located data is obtained and normalized to get the noise possibility of this data.

[0033] The specific calculation model of the noise possibility is: , where, represents the noise possibility of the i-th data in the c-th type of full charge data, represents the number of co-located data corresponding to the i-th data, represents the value of the i-th data in the c-th type of full charge data, represents the value of the j-th co-located data among the co-located data corresponding to the i-th data in the c-th type of full charge data; represents the absolute value of the difference between the i-th data and its corresponding co-located data, represents the average value of the differences between the i-th data and its data at the same cycle position (co-located data). The larger this average value is, the more the i-th data deviates from the normal data variation law, that is, the more likely it is to be noise data. Thus, the noise possibility of each data can be obtained.

[0034] The data initial segmentation unit is used to obtain the turning moment possibility of a piece of data based on the slope values of the two adjacent data corresponding to this data and the noise possibility of this data; the full charge data is segmented according to the turning moment possibilities of the data at the same moment in different types of full charge data to obtain the initial data segments.

[0035] Since data imputation using the LSTM algorithm predicts missing data based on the changing patterns of historical data, and the changing patterns of data under different working states of the circuit also vary, the accuracy of predicting missing data using historical data with a working state similar to that of the circuit at the moment corresponding to the missing data is higher. Since it is necessary to calculate the likelihood that historical data and the current missing data are in a similar working state based on the similarity between the changing patterns of data in the time period when the historical data is located and the changing patterns of data in the time period when the missing data is located, it is necessary to first divide the historical data into different time periods, and each time period should ensure the same working state. Also, since data in different states may have similar manifestations in a short period of time, while the overall changes are very different, in order to avoid misidentifying data in different working states as the same, it is also necessary to try to cover the complete data change cycle when dividing the time periods. Since the changing pattern of data will also change correspondingly when the circuit changes, the closer the historical data and the missing data are in time, the smaller the possibility of circuit change, that is, the more likely they are in the same data changing pattern. Therefore, based on the degree of proximity of the changing patterns of data in the time periods when the historical data and the current missing data are located, combined with the size of the time interval between the historical data and the current missing data, calculate the likelihood that each historical data and the current missing data are in a similar working state.

[0036] Since the changing trend of data under the same working state is relatively stable, the slope value of each data in its corresponding time-domain graph is obtained (the slope value at the position of the i-th data in the time-domain graph of the c-th full-power data). When the slope value changes significantly, it indicates that the changing trend of the data at that moment has changed significantly, that is, the working state of the circuit may have changed. Thus, calculate the likelihood that the corresponding moment of each data is the moment of working state change. Considering that the slope of the data before and after the noise data will also change significantly, it is necessary to exclude the historical data with a relatively high likelihood of being noise data.

[0037] Furthermore, based on the slope values corresponding to two adjacent data of a data and the noise likelihood of this data, obtain the likelihood of the turning moment of this data. Specifically, within a kind of full-power data, obtain the corresponding time-domain graph of this kind of full-power data, and obtain the absolute value of the difference between the slopes corresponding to two adjacent data before and after a data in the time-domain graph and normalize it to obtain the slope difference; multiply the slope difference by the difference between the preset value and the noise likelihood of this data to obtain the likelihood of the turning moment of this data.

[0038] The specific calculation model of the likelihood of the turning moment is: , Among them, is the likelihood of the turning moment of the i-th data in the c-th full-power data, indicating the likelihood that the corresponding moment of the i-th data in the c-th full-power data is the moment of working state change, represents the slope value corresponding to the data at the next moment after the i-th data, represents the slope value corresponding to the data at the previous moment before the i-th data, represents the difference value of the change trend of the i-th data before and after the corresponding moment of the data. The greater the difference, the more likely it is to be the moment of the working state change. and respectively represent the maximum and minimum values of the slope values corresponding to each data in the c-th full power data, and are used to normalize, represents the noise possibility of the i-th data, indicates that the higher the possibility that the i-th data is a noise data, the lower the possibility that the corresponding moment of the data is the real working state change moment.

[0039] Because when the working state changes, it often affects the change laws of data in multiple dimensions. In order to avoid identifying an abnormal fluctuation of a certain data as the moment of the working state change, it is necessary to comprehensively consider the possibilities that the corresponding moments of data in different dimensions are the moments of the working state change.

[0040] According to the turning moment possibilities of the data at the same moment in different types of full power data, the full power data is segmented to obtain the initial data segments. Specifically, calculate the average value of the turning moment possibilities of the data at the same moment in different types of full power data, and record it as the average turning possibility of this moment; if the average turning possibility of a moment is greater than the second reference threshold, then this moment is a suspected working state change moment, and use the suspected working state change moment to segment the full power data to obtain the initial data segments.

[0041] The calculation model of the average turning possibility is specifically as follows: , where, represents the average turning possibility of the moment corresponding to the i-th data (the i-th moment), represents the number of types of full power data, represents the turning moment possibility of the i-th data in the c-th full power data. The second reference threshold is only a reference value, and the implementer needs to adjust it according to the actual situation. After segmenting the full power data according to the suspected working state change moment, an initial data segment corresponds to the data of multiple types of full power data in the time period corresponding to this initial data segment.

[0042] The data segment correction unit is used to calculate the periodic law similarity according to the time length difference and data difference between two initial data segments; merge the initial data segments according to the periodic law similarity between every two initial data segments to obtain the final data segments.

[0043] Since the data monitored by the circuit usually changes periodically, dividing the data into different data segments only according to the moment when the data change trend changes may separate the continuous data in the same working state, which is not conducive to analyzing the overall change law of the data in different working states. Therefore, it is also necessary to merge the initial data segments obtained in the above process so that each segment in the final division result can more completely cover the monitored data in the current working state.

[0044] According to the characteristics of the data periodic change, by calculating the similarity of the data change trend and the similarity of the corresponding time length of the data in each initial data segment and other initial data segments, that is, calculating the periodic law similarity according to the time length difference and data difference between two initial data segments.

[0045] Specifically, add the absolute value of the time length difference between two initial data segments to the hyperparameter and take the reciprocal to obtain the time length similarity eigenvalue; obtain the absolute value of the difference of a certain total power data at a moment between two initial data segments, which is recorded as the difference of this total power data at this moment, sum the absolute values of the differences of this total power data at every two adjacent moments and then add the hyperparameter and take the reciprocal to obtain the data similarity eigenvalue corresponding to the two initial data segments for this total power data, and sum the data similarity eigenvalues corresponding to the two initial data segments for each total power data to obtain the comprehensive data similarity eigenvalue between the two initial data segments; multiply the time length similarity eigenvalue and the comprehensive data similarity eigenvalue and normalize them to obtain the periodic law similarity between the two initial data segments.

[0046] The calculation model of the periodic law similarity is specifically: , Among them, is the periodic law similarity between the u-th initial data segment and the v-th initial data segment, indicating the possibility that the u-th initial data segment and the v-th initial data segment are in the same periodic change law, represents the time length corresponding to the u-th initial data segment, represents the time length corresponding to the v-th initial data segment, represents the degree of proximity of the time lengths of the u-th initial data segment and the v-th initial data segment, and ε represents the hyperparameter, which is used to ensure the meaningfulness of the fraction, is the time length similarity eigenvalue, where represents the number of data contained in the u-th initial data segment, represents the absolute value of the difference between the data in the c-th total power data corresponding to the i-th moment in the u-th initial data segment and the v-th initial data segment, that is, the difference of the c-th total power data at the i-th moment, Denote the absolute value of the difference between the data in the \(u\)-th initial data segment and the \(v\)-th initial data segment corresponding to the \(c\)-th full power data at the \((i + 1)\)-th moment, that is, the difference of the \(c\)-th full power data at the \((i + 1)\)-th moment. It should be noted that if the two initial data segments are of unequal length, the misaligned positions are padded with 0. Denote the absolute value of the difference between the differences of the \(c\)-th full power data at two adjacent moments (the \(i\)-th moment and the \((i + 1)\)-th moment) (the smaller the absolute value, the closer). Sum it and add the hyperparameter, and then take the reciprocal to get , which represents the data similarity eigenvalue between the two initial data segments corresponding to the \(c\)-th full power data. The closer the difference between adjacent position data is, the more similar the data change trends of these two initial data segments are, that is, the more likely they are in the same periodic change law.

[0047] Furthermore, merge the initial data segments according to the periodic law similarity between every two initial data segments to obtain the final data segment. Set the third reference threshold and the time interval threshold. Starting from the first initial data segment, obtain the initial data segments whose periodic law similarity with the first initial data segment is greater than the third reference threshold, and denote them as suspected similar data segments; arrange the suspected similar data segments in chronological order to obtain the first sequence, and obtain the time interval between the starting moments of two adjacent suspected similar data segments in the first sequence as the adjacent interval between two adjacent suspected similar data segments; if the adjacent interval between every two adjacent suspected similar data segments in the first sequence is less than or equal to the time interval threshold, then all the suspected similar data segments in the first sequence are the periodic data segments corresponding to the first initial data segment; if there are adjacent suspected similar data segments with an adjacent interval greater than the time interval threshold, denote the adjacent interval between such adjacent suspected similar data segments as the discontinuous interval, locate the first discontinuous interval in chronological order, and at this time, the suspected similar data segments before the first discontinuous interval are the periodic data segments corresponding to the first initial data segment; and so on, starting from the first initial data segment in the remaining initial data segments in chronological order, find the periodic data segments corresponding to the first initial data segment in the remaining initial data segments until all the periodic data segments are obtained; merge the two adjacent periodic data segments that are periodic data segments in the initial data segments in chronological order to obtain the final data segment.

[0048] It should be noted that the reference values of the third reference threshold and the time interval threshold are 0.6 and 5, and the time interval threshold is 5 moments. Among all the initial data segments, some initial data segments are not periodic data segments. At this time, these data segments are not merged and directly used as the final data segments. It is also possible to have individual periodic data segments. At this time, they are not merged either and directly used as the final data segments. If there are three consecutive periodic data segments, merge the first two in order, and the last one is not merged and directly used as the final data segment.

[0049] Thus, the final data segment division result can be obtained.

[0050] A missing data completion unit is used to set a sliding window based on the length of the final data segment where a missing data is located within a full power data, and use the sliding window to slide on other final data segments to obtain the influence weights of each data within other final data segments; and complete the missing data based on each data within other final data segments and its influence weights.

[0051] According to the final data segment division result obtained by the above steps, compare the data change characteristics of the final data segment where the missing data is located with those of other final data segments. The higher the coincidence degree between the final data segment where the missing data is located and other final data segments, the higher the consistency of the working state of the circuit of the final data segment and the final data segment where the missing data is located, that is, the higher the influence degree of the historical data therein on the interpolation result of the current missing data. Set a sliding window with the time length of the final data segment where the missing data is located, and the time length of the sliding window is the same as the time length of the final data segment where the missing data is located. Slide the sliding window in each final data segment, and calculate the coincidence degree between the final data segment where the missing data is located and the data within the sliding window in turn. The maximum value of the coincidence degree is the coincidence degree between the data segment to be analyzed and the data segment where the missing data is located.

[0052] However, if a data is noise data, even if the final data segment where it is located has a high coincidence degree with the final data segment where the missing data is located, this data still cannot effectively reflect the correct interpolation result of the missing data. Therefore, considering the possibility that the data is noise and the coincidence degree between the final data segment where the data is located and the final data segment where the current missing data is located, calculate the influence degree of each data on the interpolation result of the current missing data.

[0053] Use the sliding window to slide on other final data segments to obtain the influence weights of each data within other final data segments; specifically, use the sliding window to slide on the final data segment to which a data belongs, with a sliding step of a preset value, obtain the average value of the absolute values of the differences between each data within a sliding window and the corresponding data of the final data segment to which a missing data belongs, and take the reciprocal to obtain the coincidence degree corresponding to this sliding window; obtain the maximum value among the coincidence degrees corresponding to each sliding window as the maximum coincidence degree; multiply the maximum coincidence degree by the difference between the preset value and the noise possibility of this data to obtain the influence weight of this data.

[0054] The calculation model of the influence weight of the data is specifically: , where It is the influence degree of the $i$-th data in the $c$-th total power data on the interpolation result of a missing data in the $c$-th total power data, that is, the influence weight of the $i$-th data in the $c$-th total power data. It is the noise possibility of the $i$-th data in the $c$-th total power data, and 1 is a preset value. It represents the $j$-th data in the $w$-th sliding window when the sliding window slides on the final data segment to which the $i$-th data belongs. It represents the $j$-th data in the final data segment where the missing data is located. It represents the coincidence degree of the change between the data in the sliding window and the data in the final data segment where the missing data is located. The smaller the absolute value of this difference, the higher the coincidence degree, that is, the closer the circuit working state is. The average coincidence degree is used to represent the coincidence degree of the data in the sliding window and the data in the final data segment where the current missing data is located. max represents the operation of taking the maximum value. Taking the maximum value of the coincidence degrees corresponding to each sliding window represents the coincidence degree of the data change between the final data segment where the $i$-th data is located and the final data segment where the missing data is located.

[0055] Furthermore, obtain the influence weight of each data in other final data segments except the data in the final data segment where a missing data is located. Multiply the influence weight of each data in other final data segments by each data in other final data segments to obtain the updated data corresponding to each data in other final data segments; calculate the completion result of the missing data according to the updated data corresponding to each data in other final data segments and the LSTM algorithm. Since the data values used for analysis are values after normalization processing, the interpolation result also needs to be restored according to the range of the corresponding data category.

[0056] In summary, by segmentally comparing various historical total power data, this application obtains the data change rules in different working states, and then adjusts the influence weight of historical data on the interpolation result of the missing value according to the similarity between the historical data and the working state of the missing data, as well as the time proximity between the historical data and the missing data. Finally, using the LSTM algorithm and combining the influence weight of historical data, the interpolation result at the missing value position is calculated, which can effectively solve the above problems, improve the accuracy of the collected data, and is beneficial to further analyzing the abnormal conditions of the circuit.

[0057] It should be noted that: the above sequence of the embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. And the above description of specific embodiments of this specification is given. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0058] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the differences between each embodiment and other embodiments are emphasized.

[0059] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A semi-solid circuit breaker with an intelligent full-power data acquisition function, characterized in that, The semi-solid state circuit breaker includes: A data acquisition unit, configured to respectively acquire full power data of different types within a preset time period; A noise possibility analysis unit, configured to obtain the co-located data corresponding to each data from the data within a window centered on each data; calculate the noise possibility of each data according to the co-located data corresponding to one data; A data initial segmentation unit, configured to obtain the turning moment possibility of a data based on the slope values of two adjacent data corresponding to the data and the noise possibility of the data; segment the full power data according to the turning moment possibilities of the data at the same moment in different types of full power data to obtain initial data segments; A data segmentation correction unit, configured to calculate the periodic law similarity according to the time length difference and data difference between two initial data segments; merge the initial data segments according to the periodic law similarity between every two initial data segments to obtain final data segments; A missing data completion unit, configured to set a sliding window based on the length of the final data segment where a missing data is located within one type of full power data, and slide the sliding window on other final data segments to obtain the influence weights of each data in other final data segments; complete the missing data based on each data in other final data segments and their influence weights.

2. The semi-solid circuit breaker with the intelligent full-power data acquisition function according to claim 1, wherein, The obtaining the co-located data corresponding to each data from the data within a window centered on each data includes: In one type of full power data, obtain the sum of the absolute values of the differences between the data other than the data itself within the window corresponding to one data and the data at the corresponding positions within the window of another data, then add it to a hyperparameter and take the reciprocal to obtain a reciprocal result, and normalize the reciprocal result to obtain the possibility that the other data is the data at the same position in different periods of this data; if the possibility that the other data is the data at the same position in different periods of this data is greater than a first reference threshold, then the other data is the co-located data corresponding to this data.

3. The semi-solid circuit breaker with the intelligent full-power data acquisition function according to claim 1, characterized in that, The calculating the noise possibility of a data according to the co-located data corresponding to the data includes: Obtain the average value of the absolute values of the differences between a data and its corresponding co-located data and normalize it to obtain the noise possibility of this data.

4. A semi-solid circuit breaker with an intelligent full-power data acquisition function according to claim 1, characterized in that, The obtaining the turning moment possibility of a data based on the slope values of two adjacent data corresponding to the data and the noise possibility of the data includes: Within one type of full power data, obtain the time domain diagram corresponding to this type of full power data, obtain the absolute value of the difference between the slopes of two adjacent data before and after a data in the time domain diagram and normalize it to obtain a slope difference; multiply the slope difference by the difference between a preset value and the noise possibility of this data to obtain the turning moment possibility of this data.

5. The semi-solid circuit breaker with the intelligent full-power data acquisition function according to claim 1, characterized in that, The segmenting the full power data according to the turning moment possibilities of the data at the same moment in different types of full power data to obtain initial data segments includes: Calculate the average value of the turning moment possibilities of the data at the same moment in different types of full power data, denoted as the average turning possibility at this moment; if the average turning possibility at a moment is greater than a second reference threshold, then this moment is a suspected working state change moment, and use the suspected working state change moment to segment the full power data to obtain initial data segments.

6. The semi-solid circuit breaker with the intelligent full-power data acquisition function according to claim 1, characterized in that, Calculating the periodic law similarity according to the time length difference and data difference between two initial data segments includes: Adding the absolute value of the time length difference between two initial data segments to a hyperparameter and taking the reciprocal to obtain the time length similarity eigenvalue; obtaining the absolute value of the difference between the full power data of one type at a moment between two initial data segments, which is denoted as the difference of the full power data of this type at this moment. Summing the absolute values of the differences between the differences of the full power data of this type at every two adjacent moments and adding a hyperparameter, then taking the reciprocal to obtain the data similarity eigenvalue corresponding to the two initial data segments for this type of full power data. Summing the data similarity eigenvalues corresponding to each type of full power data between the two initial data segments to obtain the comprehensive data similarity eigenvalue between the two initial data segments; multiplying the time length similarity eigenvalue and the comprehensive data similarity eigenvalue and normalizing to obtain the periodic law similarity between the two initial data segments.

7. The semi-solid circuit breaker with the intelligent full-power data acquisition function according to claim 1, characterized in that, Merging the initial data segments according to the periodic law similarity between every two initial data segments to obtain the final data segment includes: Starting from the first initial data segment, obtaining the initial data segments whose periodic law similarity with the first initial data segment is greater than the third reference threshold, which are denoted as suspected similar data segments; arranging the suspected similar data segments in chronological order to obtain the first sequence, and obtaining the time interval between the starting moments of two adjacent suspected similar data segments in the first sequence as the adjacent interval between two adjacent suspected similar data segments; if the adjacent interval between every two adjacent suspected similar data segments in the first sequence is less than or equal to the time interval threshold, then all the suspected similar data segments in the first sequence are the periodic data segments corresponding to the first initial data segment; if there are adjacent suspected similar data segments with an adjacent interval greater than the time interval threshold, denoting the adjacent interval between such adjacent suspected similar data segments as the discontinuous interval, locating the first discontinuous interval in chronological order, and at this time, the suspected similar data segments before the first discontinuous interval are the periodic data segments corresponding to the first initial data segment; and so on, starting from the first initial data segment in the remaining initial data segments in chronological order, finding the periodic data segments corresponding to the first initial data segment in the remaining initial data segments until all the periodic data segments are obtained; merging two adjacent periodic data segments that are periodic data segments in the initial data segments in chronological order to obtain the final data segment.

8. The semi-solid circuit breaker with the intelligent full-power data acquisition function according to claim 1, characterized in that, Using a sliding window to slide on other final data segments to obtain the influence weights of each data in other final data segments includes: Using a sliding window to slide on the final data segment to which a data belongs, with a sliding step size of a preset value, obtaining the mean of the absolute values of the differences between the data in a sliding window and the corresponding data in the final data segment to which a missing data belongs, and taking the reciprocal to obtain the coincidence degree corresponding to this sliding window; obtaining the maximum value among the coincidence degrees corresponding to each sliding window as the maximum coincidence degree; multiplying the maximum coincidence degree by the difference between the preset value and the noise possibility of this data to obtain the influence weight of this data.

9. A semi-solid circuit breaker with an intelligent full-power data acquisition function according to claim 1, characterized in that, Completing the missing data based on each data and its influence weight in other final data segments, including: Obtaining the influence weight of each data in other final data segments except the data in the final data segment where a missing data is located, multiplying the influence weight of each data in other final data segments by each data in other final data segments respectively to obtain the updated data corresponding to each data in other final data segments; Completing the missing data according to the updated data corresponding to each data in other final data segments and the LSTM algorithm.

Images