Circuit breaker operation status monitoring method and system based on big data

Abstract

The present invention relates to the technical field of data processing, and particularly relates to a method and system for monitoring the operating state of a circuit breaker based on big data, including: obtaining an operating state vector; using a filtering algorithm to obtain the gain influence at each acquisition moment, and determining a fluctuation eigenvalue based on the curve fitting result in the data space constructed based on the gain influence; determining the fluctuation degree of the monitoring data at the acquisition moment according to the influence degree of the data fluctuation at the data anomaly moment on the monitoring data at the acquisition moment and the fluctuation eigenvalue, and completing the update of the Kalman gain matrix at the acquisition moment; using an extended Kalman filtering algorithm to obtain a prediction result of the operating state vector based on the updated Kalman gain matrix, and obtaining a monitoring result of the operating state of the circuit breaker based on the prediction result. The present invention improves the accuracy of monitoring the change of the operating state of the circuit breaker by more accurately predicting the monitoring data of the operating state of the circuit breaker.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a method and system for monitoring the operating state of a circuit breaker based on big data. Background Art

[0002] As an important power equipment in the power system, the safe and stable operating state of a circuit breaker is the premise for the stable operation of the power system. Therefore, it is particularly crucial to monitor the state of the circuit breaker. Usually, sensors are used to monitor the working environment parameters and electrical parameters of the circuit breaker, and the operating state of the circuit breaker is judged by analyzing various data.

[0003] Wireless sensor network technology is a circuit breaker state monitoring technology widely used in recent years. By deploying various sensors around the circuit breaker, a large amount of monitoring data is transmitted to the monitoring terminal through wireless communication, and real-time monitoring of the operating state of the circuit breaker is achieved by using big data analysis technology.

[0004] At the monitoring terminal, the existing technology uses the extended Kalman filter (EKF) algorithm to predict and analyze the historical monitoring data of the circuit breaker, judge whether the operating state of the circuit breaker is abnormal, and view the abnormal state in advance to ensure the safety of the power system. When analyzing the historical monitoring data of the circuit breaker to judge the current operating state, the EKF algorithm will cover up the outlier fluctuations caused by local abnormal change data due to the long-term stable operation situation, resulting in a large deviation of the predicted value from the actual situation when the local data selection range is inappropriate, thus reducing the accuracy of abnormal monitoring of the operating state of the circuit breaker. Summary of the Invention

[0005] To solve the above problems, the present invention provides a method and system for monitoring the operating state of a circuit breaker based on big data.

[0006] In a first aspect, an embodiment of the present invention provides a method for monitoring the operating state of a circuit breaker based on big data, the method comprising the following steps:

[0007] Deploy sensors around the circuit breaker to obtain the operating state vector at each acquisition moment;

[0008] Use a filtering algorithm to determine the gain influence at each acquisition moment based on the operating state vector at each acquisition moment, and determine the fluctuation characteristic value at each acquisition moment based on the curve fitting result in the data space constructed based on the gain influence;

[0009] According to the influence degree of the data fluctuation at the data anomaly moment determined by the anomaly points in each type of monitoring data during the collection period on each type of monitoring data at each collection moment, and the fluctuation characteristic value of each collection moment, determine the fluctuation degree of each type of monitoring data at each collection moment; use the fluctuation degrees of all types of monitoring data at each collection moment to construct the updated Kalman gain matrix at each collection moment;

[0010] Use the extended Kalman filter algorithm to obtain the prediction result of the operating state vector at each collection moment based on the updated Kalman gain matrix at each collection moment, and obtain the monitoring result of the circuit breaker operating state based on the prediction result of the operating state vector at each collection moment.

[0011] In one embodiment, the determination of the gain influence at each collection moment includes the following specific method:

[0012] Take the operating state vector at each collection moment as the input, and use the extended Kalman filter algorithm to respectively obtain the Kalman gain matrix and the first prediction vector at each collection moment, where the first prediction vector is the prediction result of the operating state vector;

[0013] Determine the predictability of the operating state vector at each collection moment based on the change amount of the similarity degree between the operating state vectors at each collection moment and the adjacent previous moment and the first prediction vector;

[0014] Take the accumulated result of the product of the operating state vector at each collection moment and the column vectors in the Kalman gain matrix at each collection moment as the numerator;

[0015] Take the sum of the predictability of the operating state vector at each collection moment and the constant parameter as the denominator;

[0016] Take the ratio of the numerator to the denominator as the gain influence at each collection moment.

[0017] In one embodiment, the determination of the predictability of the operating state vector at each collection moment includes the following specific method:

[0018] Take the absolute value of the difference between the cosine similarity between the operating state vectors at each collection moment and the adjacent previous moment and the cosine similarity between the first prediction vectors at each collection moment as the input of the Singh function, and take the absolute value of the output function value as the predictability of the operating state vector at each collection moment.

[0019] In one embodiment, the determination of the fluctuation characteristic value at each collection moment includes the following specific method:

[0020] Construct a data space with all the acquisition times as the abscissa and the gain impacts at all the acquisition times as the ordinate, map the gain impacts at all the acquisition times into the data space to obtain a number of data points, and acquire the fitting curve of all the data points by means of curve fitting;

[0021] Determine the deviation weight of each acquisition time based on the fitting deviation of all the acquisition times on the fitting curve;

[0022] Calculate the product of the absolute value of the difference between the fitting value of each acquisition time on the fitting curve and the gain impact of each acquisition time and the deviation weight of each acquisition time as the fluctuation characteristic value of each acquisition time.

[0023] In one embodiment, the method for determining the deviation weight of each acquisition time specifically includes:

[0024] Calculate the ratio of the fitting deviation of each acquisition time on the fitting curve to the sum of the fitting deviations of all the acquisition times on the fitting curve as the deviation weight of each acquisition time.

[0025] In one embodiment, the method for determining the fluctuation degree of each type of monitoring data at each acquisition time specifically includes:

[0026] Take all the acquisition times of each type of monitoring data within the acquisition period as inputs, use the method of anomaly detection to determine all the anomaly points in each type of monitoring data, and take the acquisition time corresponding to each anomaly point as a data anomaly time;

[0027] Determine the anomaly impact weight of each acquisition time based on the degree of influence of the data fluctuation at the adjacent data anomaly times on each acquisition time;

[0028] Take a preset number of adjacent acquisition times on the left and right of each acquisition time as the neighboring times centered on each acquisition time, and calculate the discrete characteristic index of each type of monitoring data at each acquisition time and all the neighboring times as the first characteristic value of each type of monitoring data at each acquisition time;

[0029] Take the product of the ratio of the first characteristic value of each type of monitoring data at each acquisition time to the distribution variance of each type of monitoring data within the acquisition period, the anomaly impact weight of each acquisition time, and the fluctuation characteristic value of each acquisition time as the fluctuation degree of each type of monitoring data at each acquisition time.

[0030] In one embodiment, the method for determining the anomaly impact weight of each acquisition time specifically includes:

[0031] For each data anomaly moment, all neighboring moments centered on each data anomaly moment are arranged in ascending order of the values of each type of monitored data, obtaining the local monitored data sequence for each data anomaly moment;

[0032] Calculate the difference measurement results between the local monitored data sequence of the data anomaly moment with the shortest time interval from each acquisition moment and the local monitored data sequences of the remaining data anomaly moments, and take the mean of the cumulative results of the difference measurement results over all data anomaly moments as the anomaly influence weight for each acquisition moment.

[0033] In one embodiment, the specific method for constructing the updated Kalman gain matrix for each acquisition moment includes:

[0034] For any acquisition moment, linearly normalize the fluctuation degrees of all monitored data in the operation state vector of each acquisition moment, and take each normalized fluctuation degree as a fluctuation evaluation factor;

[0035] Take all the fluctuation evaluation factors at each acquisition moment as the main diagonal elements in the fluctuation evaluation factor matrix of each acquisition moment; fill all elements other than the main diagonal in the fluctuation evaluation factor matrix of each acquisition moment with 0, obtaining the fluctuation evaluation factor matrix of each acquisition moment;

[0036] Take the product of the fluctuation evaluation factor matrix of each acquisition moment and the Kalman gain matrix of each acquisition moment as the updated Kalman gain matrix of each acquisition moment.

[0037] In one embodiment, the specific method for obtaining the monitoring result of the circuit breaker operation state based on the prediction result of the operation state vector at each acquisition moment includes:

[0038] Input the operation state vector of each acquisition moment and the updated Kalman gain matrix of each acquisition moment into the extended Kalman filter algorithm to obtain the predicted operation state vector of each acquisition moment;

[0039] Calculate the absolute value of the difference between the same type of monitored data in the operation state vector and the predicted operation state vector at each acquisition moment, and take the ratio of the absolute value of the difference to the same type of monitored data in the operation state vector as the deviation ratio of the same type of monitored data;

[0040] Compare the deviation ratio of each type of monitoring data with a preset warning threshold respectively. When the deviation ratio of any type of monitoring data is greater than or equal to the warning threshold, it is considered that the operating state of the circuit breaker has changed significantly, and the acquisition results and deviation ratios of all types of monitoring data are sent to the circuit breaker maintenance personnel at the same time; when the deviation ratios of all types of monitoring data are less than the warning threshold, it is considered that the operating state of the circuit breaker has not changed.

[0041] In a second aspect, the embodiment of the present application further provides a circuit breaker operating state monitoring system based on big data, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above-mentioned circuit breaker operating state monitoring method based on big data.

[0042] The beneficial effects of the technical solution of the present invention are:

[0043] According to the fluctuation eigenvalue and the change of each type of monitoring data in the operation state vector at each acquisition moment under all corresponding historical moments, the present invention obtains the fluctuation eigenvalue at each acquisition moment; the fluctuation eigenvalue reflects the difference degree of the operation state vector at each moment during data prediction compared with the operation state vectors at other acquisition moments; then, using the characteristic that the same abnormal state change will not occur repeatedly in a short time within the period of actually monitoring the operation state of the circuit breaker, resulting in different local fluctuations of the data at the data abnormal moment, the present invention determines the fluctuation degree of each type of monitoring data at each acquisition moment, and uses the fluctuation degree to complete the update of the Kalman gain matrix, so that when the EKF algorithm analyzes the historical monitoring data of the circuit breaker to judge the current operation state, it has an impact on the outlier fluctuations generated by the local abnormal change data, improves the accuracy of the operation state vector prediction result, and further improves the accuracy of the circuit breaker operation state monitoring result. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for 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, without creative efforts, other drawings can be obtained based on these drawings.

[0045] Figure 1 It is a flowchart of the steps of the circuit breaker operating state monitoring method based on big data of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following will, in conjunction with the accompanying drawings and preferred embodiments, elaborate in detail on the specific implementation manner, structure, features, and effects of the method and system for monitoring the operating state of a circuit breaker based on big data 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.

[0047] 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.

[0048] The following will specifically describe the specific solutions of the method and system for monitoring the operating state of a circuit breaker based on big data provided by the present invention in conjunction with the accompanying drawings.

[0049] Please refer to Figure 1 , which shows a flowchart of the steps of the method for monitoring the operating state of a circuit breaker based on big data provided by an embodiment of the present invention. The method includes the following steps:

[0050] Step S001: Obtain the operating state vectors at a plurality of acquisition moments during the operation of the circuit breaker.

[0051] The present invention aims to analyze the monitoring data collected by multiple sensors during the operation of the circuit breaker to determine whether the operating state of the circuit breaker is normal. When analyzing the monitoring data using the EKF algorithm, the Kalman gain matrix at each acquisition moment is calculated by analyzing the change characteristics of the monitoring data at adjacent acquisition moments and the change situation of each monitoring data at all acquisition moments. Based on the Kalman gain matrix, the operating state of the circuit breaker is predicted to determine the moment when the operating state is abnormal. Therefore, the present application first needs to deploy sensors to collect a large amount of monitoring data.

[0052] Specifically, the sensors are deployed around the circuit breaker to ensure that the frequency of collecting monitoring data by the sensors is consistent and the acquisition duration is the same. Among them, the sensors include, but are not limited to, voltage sensors, current sensors, temperature sensors, and humidity sensors. In an embodiment of the present invention, the acquisition period is 5 minutes and the data acquisition frequency is 100HZ. In other embodiments, the acquisition period and acquisition frequency can be set by the implementer, and the present application does not make special restrictions on this. For any acquisition moment, the vector composed of the monitoring data collected by all sensors at each acquisition moment is used as the operating state vector at each acquisition moment.

[0053] Step S002: Obtain the gain influence at each acquisition moment according to the difference in the operation state vectors between each acquisition moment and the adjacent acquisition moments; obtain the fluctuation characteristic value at each acquisition moment according to the difference situation of the gain influence between each acquisition moment and the adjacent acquisition moments.

[0054] During the operation of the circuit breaker, different operation states of the circuit breaker will lead to different change situations of the same monitoring data at different acquisition moments. Therefore, for the fluctuation situation of each monitoring data in the operation state vector, the present invention combines the deviation situation between the predicted value and the observed value caused by the change of the Kalman gain between the acquisition moments to obtain the gain influence at each acquisition moment.

[0055] Degree of fluctuation of monitoring dataGain influenceSpecifically, for any acquisition moment, taking the t-th acquisition moment as an example, input the operation state vector of the t-th acquisition moment into the extended Kalman filter algorithm to obtain the Kalman gain matrix and the first prediction vector at the t-th acquisition moment, and the first prediction vector is the prediction result of the operation state vector; obtain the gain influence at the t-th acquisition moment according to the difference in the operation state vectors between the t-th acquisition moment and the adjacent acquisition moments. Among them, the extended Kalman filter algorithm is a prior art and will not be elaborated here in this embodiment.

[0056] As an example, the calculation method for obtaining the gain influence at the t-th acquisition moment is as follows:

[0057]

[0058] In the formula, represents the gain influence at the t-th acquisition moment; represents the total number of all column vectors in the Kalman gain matrix at the t-th acquisition moment; represents the j-th column vector in the Kalman gain matrix at the t-th acquisition moment; represents the operation state vector at the t-th acquisition moment; represents the operation state vector at the (t - 1)-th acquisition moment; represents the first prediction vector at the t-th acquisition moment; represents the first prediction vector at the (t - 1)-th acquisition moment; represents the Singh function; represents taking the absolute value; represents the cosine similarity between the operation state vector at the t-th acquisition moment and the operation state vector at the (t - 1)-th acquisition moment; represents the cosine similarity between the first prediction vector at the t-th acquisition moment and the first prediction vector at the (t - 1)-th acquisition moment.

[0059] Among them, the numerator It reflects the deviation of the monitoring data at the t-th acquisition moment. Multiply all column vectors in the Kalman gain matrix at the t-th moment of the operating state vector by the operating state vector at the t-th acquisition moment and accumulate them. The larger the value of the numerator, the greater the difference in the monitoring data at the t-th acquisition moment compared with the adjacent moments, and the greater the change in the operating state of the circuit breaker at the t-th acquisition moment compared with the operating state of the circuit breaker at the adjacent acquisition moments. The denominator reflects the predictability of the operating state vector at the t-th acquisition moment, which is judged by the similarity between the change in the predicted value at the previous moment and the change in the observed value at the t-th acquisition moment. The larger the denominator, the closer the change in the predicted value is to the change in the observed value, and the smaller the possibility of deviation in the operating state of the circuit breaker at the t-th acquisition moment compared with the (t - 1)-th acquisition moment, and the stronger the predictability.

[0060] It should be further noted that the magnitude of the gain impact reflects the deviation of the predicted value from the actual observed value at each acquisition moment. However, when there are noise points in the acquisition process of multiple sensors, the noise points will affect the predicted value and the observed value at a certain acquisition moment, resulting in the calculated gain impact not conforming to the actual situation. Therefore, in order to avoid the influence of noise points on the predicted value during the data acquisition process, the gain impact at each acquisition moment is weighted and judged according to the gain impact at adjacent acquisition moments to obtain the fluctuation characteristic value at each acquisition moment.

[0061] Specifically, taking the acquisition moment as the abscissa and the gain impact as the ordinate to construct a two-dimensional coordinate system, map the gain impact at each acquisition moment into the two-dimensional coordinate system to obtain an acquisition data point, and obtain several acquisition data points, and obtain the fitting curve of all acquisition data points by curve fitting. And calculate the fitting deviation of each acquisition moment on the fitting curve respectively.

[0062] Among them, curve fitting is a commonly used technique in the field of data analysis, and the specific process is not described in this application. Commonly used curve fittings include but are not limited to curve fitting based on local weighted regression, least squares fitting, and polynomial fitting. This application does not limit the specific method of curve fitting. Preferably, as an embodiment of this application, the least squares fitting method is used to obtain the fitting curve.

[0063] First, calculate the deviation weight at each acquisition moment according to the magnitude of the fitting deviation at each acquisition moment, which is used to characterize the outlier situation of the distribution of the gain impact at each acquisition moment compared with the other moments. The calculation process of the deviation weight at the t-th acquisition moment is as follows:

[0064]

[0065] In the formula, is the deviation weight at the t-th acquisition moment, is the fitting deviation at the t-th acquisition moment; m is the number of acquisition moments.

[0066] After that, the fluctuation eigenvalue of each acquisition moment is determined by combining the magnitude of the fitting deviation of each acquisition moment. The larger the fluctuation eigenvalue, the more significant the fluctuation difference between the data fitting result and the gain influence at this acquisition moment, and the greater the difference between the operation state vector at this moment and the operation state vectors at other acquisition moments.

[0067]

[0068] In the formula, is the fluctuation eigenvalue at the t-th acquisition moment, is the absolute value of the difference between the fitting value of the t-th acquisition moment on the fitting curve and the gain influence of the t-th acquisition moment.

[0069] Step S003: Obtain the fluctuation degree of each type of monitoring data at each acquisition moment according to the abnormal points of each type of monitoring data within the acquisition period and the influence degree of the data fluctuation at the data abnormal moment on each type of monitoring data; obtain the updated Kalman gain matrix at each acquisition moment according to the fluctuation degree.

[0070] It should be noted that in the operation state vector of each acquisition moment, each type of monitoring data can reflect whether there is an abnormality in the operation state of the circuit breaker, but the sensitivity of different types of monitoring data to the data when the operation state of the circuit breaker is abnormal is different. For example, usually, the action voltage range of the circuit breaker is usually between 70% and 110% of the rated voltage; while the installation environment temperature range of the circuit breaker is usually -40°C to 40°C. When the operation state of the circuit breaker is abnormal, it usually shows that the change in electrical parameters is more significant than the change in temperature and humidity. Therefore, if there are large data fluctuations in a certain type of monitoring data at multiple acquisition moments, it is necessary to analyze the data fluctuations to evaluate whether it corresponds to a change in the operation state of the circuit breaker.

[0071] Furthermore, for any type of monitoring data, if there are a large number of acquisition moments with large fluctuations in the monitoring data of this type, it means that this type of monitoring data is more likely to generate large data fluctuations and there should be a large fluctuation range throughout the data acquisition period. If the fluctuation range of a certain type of monitoring data is generally small and there are only large fluctuations at the acquisition moments within a certain time period, it means that the data fluctuations within this time period are very likely to be caused by the abnormal operation state of the circuit breaker.

[0072] Specifically, for any type of monitoring data, taking the y-th type of monitoring data as an example, the abnormal detection method is used to detect the abnormal points in all the y-th type of monitoring data, and the acquisition moments corresponding to the abnormal points are marked as data abnormal moments.

[0073] It should be noted that data anomaly detection is a commonly used technology in the field of data processing, and the specific process will not be elaborated here. Commonly used data anomaly detection methods include, but are not limited to, Local Outlier Factor (LOF) detection, Isolation Forest detection, and 3-sigma criterion detection. This application does not impose special restrictions on the method of anomaly detection. Preferably, as an embodiment of this application, the anomaly points in the y-th type of monitoring data are obtained by using LOF detection.

[0074] Furthermore, for each acquisition moment, the closer the time interval is to the data anomaly moment, the more likely it is to be in the same operating state as the breaker at the data anomaly moment. Usually, when a breaker failure causes an abnormal operating state, maintenance personnel will perform targeted repairs according to the cause of the failure, and the same failure will not occur repeatedly in a short period of time. Therefore, if the local fluctuation situation of the y-th type of monitoring data at a certain data anomaly moment is inconsistent with that at the adjacent data anomaly moment, then this data anomaly moment is more likely to correspond to the change in the operating state during the breaker failure.

[0075] Specifically, with each acquisition moment as the center, k adjacent acquisition moments are taken on both the left and right as neighboring moments. The discrete characteristic index of the 2k + 1 y-th type of monitoring data at each acquisition moment and the 2k neighboring moments is calculated as the first eigenvalue of the y-th type of monitoring data at each acquisition moment. The larger the discrete characteristic index, the more complex the data fluctuation degree within the time period of the acquisition moment. Secondly, for each data anomaly moment, the 2k + 1 y-th type of monitoring data taken with each data anomaly moment as the center is arranged in ascending order of the values to obtain the local monitoring data sequence of each data anomaly moment. In this embodiment, the value of k is set to the empirical value 10. In other embodiments, the implementer can set the value of k according to the data acquisition frequency, and this application does not impose special restrictions on this.

[0076] Among them, the discrete characteristic index is used to evaluate the discrete degree of the distribution of the 2k + 1 y-th type of monitoring data. The greater the discrete degree, the larger the value of the corresponding discrete characteristic index. On the premise of being able to achieve this purpose, in different embodiments, different evaluation indexes such as information entropy, distribution variance, and coefficient of variation can be used as the discrete characteristic index.

[0077] Here, the fluctuation degree of each type of monitoring data at each acquisition moment is calculated to characterize the local fluctuation degree of each type of monitoring data at each acquisition moment compared to the distribution of the monitoring data within the acquisition period. The calculation process of the fluctuation degree of the y-th type of monitoring data at the t-th acquisition moment is as follows:

[0078]

[0079] In the formula, is the abnormal influence weight at the t-th acquisition moment, and M is the number of data abnormal moments within the acquisition period. is the local monitoring data sequence of the data abnormal moment with the shortest time interval from the t-th acquisition moment. is the local monitoring data sequence of the b-th data abnormal moment within the acquisition period. is the local monitoring data sequence 、 is the difference measurement result between them.

[0080] Among them, the difference measurement result is used to measure the difference between the local monitoring data sequences 、 The greater the difference, the greater the value of the corresponding difference measurement result. On the premise of being able to achieve this purpose, in different embodiments, different measurement results such as DTW distance, Euclidean distance, value variance, etc. can be used as the difference measurement result.

[0081] After that, in combination with the influence of the adjacent data abnormal moments on the t-th acquisition moment, determine the fluctuation degree of the y-th type of monitoring data at the t-th acquisition moment:

[0082]

[0083] In the formula, is the fluctuation degree of the y-th type of monitoring data at the t-th acquisition moment, is the first eigenvalue of the y-th type of monitoring data at the t-th acquisition moment, is the distribution variance of the y-th type of monitoring data at all acquisition moments within the acquisition period.

[0084] Among them, the smaller the overall data fluctuation range of the y-th type of monitoring data throughout the period, the more prominent the local fluctuation of the y-th type of monitoring data at the t-th acquisition moment, indicating that the operating state of the circuit breaker reflected by the y-th type of monitoring data in the local time period where the t-th acquisition moment is located is more likely to change to a greater extent; at the same time, the stronger the difference between the local monitoring data sequences of the data abnormal moment with the shortest interval and the other data abnormal moments, the more likely it is that there is a fault in the circuit breaker, resulting in a large change in the operating state, and the greater the impact on the monitoring data at the t-th acquisition moment.

[0085] Further, for any acquisition moment, the fluctuation degrees of all the monitoring data in the operation state vector at each acquisition moment are linearly normalized, and each normalized fluctuation degree is denoted as a fluctuation evaluation factor. Secondly, the fluctuation evaluation factors of all the monitoring data in the operation state vector at each acquisition moment are used as the main diagonal elements in the fluctuation evaluation factor matrix at each acquisition moment; all the elements except the main diagonal in the fluctuation evaluation factor matrix at each acquisition moment are set to 0, obtaining the fluctuation evaluation factor matrix at each acquisition moment; thereafter, the product of the fluctuation evaluation factor matrix at each acquisition moment and the Kalman gain matrix at each acquisition moment is used as the updated Kalman gain matrix at each acquisition moment.

[0086] Step S004: Perform real-time monitoring of the breaker operation state according to the updated Kalman gain matrix.

[0087] Specifically, for any acquisition moment, each acquisition moment is denoted as the target acquisition moment; the operation state vector at the target acquisition moment and the updated Kalman gain matrix at the target acquisition moment are input into the extended Kalman filter algorithm to obtain the predicted operation state vector at the target acquisition moment. Among them, the extended Kalman filter algorithm is a well-known technology, and the specific process will not be elaborated herein.

[0088] Further, to ensure the safe operation of the breaker, it should be considered that the breaker operation state has changed when there is a prediction deviation in any one of the monitoring data of the breaker operation state, and the maintenance personnel should be reminded to conduct on-site maintenance. Therefore, in this application, the absolute value of the difference between the same type of monitoring data in the operation state vector at the target acquisition moment and the predicted operation state vector is calculated respectively, and the ratio of the absolute value of the difference to the same type of monitoring data in the operation state vector is used as the deviation ratio of the same type of monitoring data, obtaining the deviation ratios of all types of monitoring data at the target acquisition moment. And they are compared with the preset warning thresholds respectively. When the deviation ratio of any one of the monitoring data is greater than or equal to the warning threshold, it is considered that the breaker operation state has changed significantly, and the acquisition results and deviation ratios of all types of monitoring data are sent to the breaker maintenance personnel at the same time; when the deviation ratios of all types of monitoring data are less than the warning threshold, it is considered that the breaker operation state has not changed.

[0089] The present invention also proposes a breaker operation state monitoring system based on big data, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the breaker operation state monitoring method based on big data described in steps S001 to S004 are implemented.

[0090] It should be noted that the above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limiting them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the present application, and should all be included within the protection scope of the present application.

[0091] The various embodiments in this specification are all described in a progressive manner. For the same or similar parts between the various embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A circuit breaker operation status monitoring method based on big data, characterized in that: The method comprises the following steps: Sensors are placed around the circuit breaker to obtain the operating state vector at each acquisition moment; Determine the gain influence at each acquisition moment based on the running state vector at each acquisition moment by using a filtering algorithm, and determine the fluctuation characteristic value at each acquisition moment based on the curve fitting result in the data space constructed based on the gain influence; Take each type of monitoring data at all collection moments in the collection cycle as input, use anomaly detection to determine all abnormal points in each type of monitoring data, and take the collection moment corresponding to each abnormal point as a data abnormal moment; determine the abnormal impact weight of each collection moment based on the degree to which each collection moment is affected by the data fluctuation at the adjacent data abnormal moment; take each collection moment as the center, take a preset number of adjacent collection moments on the left and right as neighboring moments, calculate the discrete characteristic index of each type of monitoring data at each collection moment and all neighboring moments as the first eigenvalue of each type of monitoring data at each collection moment; take the product of the ratio of the first eigenvalue of each type of monitoring data at each collection moment to the distribution variance of each type of monitoring data in the collection cycle, the abnormal impact weight of each collection moment, and the fluctuation characteristic value of each collection moment as the fluctuation degree of each type of monitoring data at each collection moment; For any collection moment, the fluctuation degree of all monitoring data in the operation state vector at each collection moment is linearly normalized, and each normalized fluctuation degree is used as a fluctuation evaluation factor; all the fluctuation evaluation factors at each collection moment are used as the main diagonal elements in the fluctuation evaluation factor matrix at each collection moment; all elements except the main diagonal in the fluctuation evaluation factor matrix at each collection moment are filled with 0 to obtain the fluctuation evaluation factor matrix at each collection moment; the product of the fluctuation evaluation factor matrix at each collection moment and the Kalman gain matrix at each collection moment is used as the updated Kalman gain matrix at each collection moment; the extended Kalman filter algorithm is used to obtain the prediction result of the operation state vector at each collection moment based on the updated Kalman gain matrix at each collection moment, and the monitoring result of the circuit breaker operation state is obtained based on the prediction result of the operation state vector at each collection moment.

2. The circuit breaker operation status monitoring method based on big data according to claim 1 is characterized in that: The specific method of determining the gain influence at each acquisition moment includes: Taking the running state vector at each acquisition moment as input, using the extended Kalman filter algorithm to respectively obtain the Kalman gain matrix and the first prediction vector at each acquisition moment, wherein the first prediction vector is the prediction result of the running state vector; Determine the predictability of the running state vector at each collection moment based on the change in the similarity between the running state vector at each collection moment and the previous moment and the first prediction vector; The accumulated result of the product of the running state vector at each acquisition moment and the column vector in the Kalman gain matrix at each acquisition moment is used as the numerator; The sum of the predictability of the operating state vector at each acquisition moment and the constant parameter is used as the denominator; The ratio of the numerator to the denominator is taken as the gain impact at each acquisition moment.

3. The circuit breaker operation status monitoring method based on big data according to claim 2 is characterized in that: The specific method of determining the predictability of the operating state vector at each acquisition moment is as follows: The absolute value of the difference between the cosine similarity between the operating state vector at each acquisition moment and the adjacent previous moment and the cosine similarity between the first prediction vector at each acquisition moment and the adjacent previous moment is used as the input of the Singer function, and the absolute value of the output function value is used as the predictability of the operating state vector at each acquisition moment.

4. The circuit breaker operation status monitoring method based on big data according to claim 1 is characterized in that: The specific method of determining the fluctuation characteristic value at each acquisition moment includes: The data space is constructed with all acquisition moments as the horizontal coordinate and the gain influence of all acquisition moments as the vertical coordinate, the gain influence of all acquisition moments is mapped into the data space to obtain a number of data points, and the fitting curve of all data points is obtained by curve fitting; Determine the deviation weight of each acquisition moment based on the fitting deviation of all acquisition moments on the fitting curve; The product of the absolute value of the difference between the fitting value on the fitting curve at each acquisition moment and the gain influence at each acquisition moment and the deviation weight at each acquisition moment is calculated as the fluctuation characteristic value at each acquisition moment.

5. The circuit breaker operation status monitoring method based on big data according to claim 4 is characterized in that: The specific method of determining the deviation weight at each acquisition moment includes: The ratio of the fitting deviation on the fitting curve at each acquisition moment to the cumulative sum of the fitting deviations on the fitting curve at all acquisition moments is calculated as the deviation weight of each acquisition moment.

6. The circuit breaker operation status monitoring method based on big data according to claim 1 is characterized in that: The specific method of determining the abnormal impact weight of each acquisition moment includes: For each data anomaly moment, all monitoring data of all neighboring moments taken around each data anomaly moment are arranged in the order of value size to obtain a local monitoring data sequence for each data anomaly moment; Calculate the difference measurement results between the local monitoring data sequence at the data anomaly moment with the shortest time interval between each acquisition moment and the local monitoring data sequences at the remaining data anomaly moments, and take the average of the accumulated results of the difference measurement results at all data anomaly moments as the anomaly impact weight of each acquisition moment.

7. The circuit breaker operation status monitoring method based on big data according to claim 1 is characterized in that: The monitoring result of the circuit breaker operating state is obtained based on the prediction result of the operating state vector at each acquisition moment, and the specific method includes: Input the running state vector at each acquisition moment and the updated Kalman gain matrix at each acquisition moment into the extended Kalman filter algorithm to obtain the predicted running state vector at each acquisition moment; Calculate the absolute value of the difference between the running state vector at each acquisition moment and the same monitoring data in the predicted running state vector, and use the ratio of the absolute value of the difference to the same monitoring data in the running state vector as the deviation ratio of the same monitoring data; The deviation ratio of each type of monitoring data is compared with the preset warning threshold value. When the deviation ratio of any type of monitoring data is greater than or equal to the warning threshold value, it is considered that the operating state of the circuit breaker has changed significantly, and the collection results and deviation ratios of all types of monitoring data are sent to the circuit breaker maintenance personnel at the same time; when the deviation ratios of all types of monitoring data are less than the warning threshold value, it is considered that the operating state of the circuit breaker has not changed.

8. A circuit breaker operation status monitoring system based on big data, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the circuit breaker operating status monitoring method based on big data as described in any one of claims 1 to 7 are implemented.

Images