Data training method based on high-voltage switch opening and closing speed detection

Abstract

The present invention discloses a data training method based on high-voltage switch opening and closing speed detection, which belongs to the field of data analysis technology and comprises the following steps: collecting data on the opening and closing speed, time and position of the high-voltage switch related to the laser and collecting data on the reflected light after diffuse reflection are stored in the form of data units, and converted into matrix tensor product data units for calculating the opening and closing speed of the high-voltage circuit breaker; performing low-rank decomposition on the converted stored data units to realize data query and analysis; further providing query results and analysis results of the data according to the requirements of speed detection; submitting the query results and analysis results to the host for decision-making, and according to the decision results after the collection data training displayed by the host, the operation and maintenance personnel decide whether to repair or replace the action mechanism of the high-voltage switch.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis training, and in particular to a data analysis method for detecting the opening and closing speed of a high-voltage switch. Background Art

[0002] With the rapid development of society, people have higher and higher requirements for the safety and reliability of electricity use. High-voltage switches have the dual tasks of control and protection in the power system, and their performance is directly related to the safe operation of the power system. Mechanical characteristic parameters are one of the important parameters for judging the performance of high-voltage switches. The opening speed of the high-voltage switch has an important influence on the breaking performance and service life of the high-voltage switch. If the opening speed of the high-voltage switch is too fast, the high-voltage switch operating mechanism and various transmission components will exceed the mechanical force they can withstand, causing damage to the buffer and shortening the service life of the circuit breaker; if the opening speed is too slow, the normal operating current or fault current cannot be quickly cut off, which will prolong the arcing time between the contacts, causing hazards such as contact burning and even arc extinguishing chamber explosion; if the closing speed is too slow, when the circuit breaker is closed on a short-circuit fault, it may not be able to overcome the effect of the contact closing electromotive force, thereby causing contact vibration or movement stagnation, and then causing contact burning or arc extinguishing chamber explosion; the speed test of high-voltage switches has long been regarded as a difficulty in the field of mechanical characteristics testing of high-voltage switches; the existing detection methods mostly use contact linear sensors, rotary sensors, and acceleration sensors for speed and stroke measurement. The host's simple calculation of the collected time, distance, and speed has large errors, and the detection accuracy needs to be improved.

[0003] In order to overcome the defects of the traditional high-voltage switch opening and closing speed detection method and realize high-precision speed detection of the high-voltage switch opening process, this method uses the data training method based on the high-voltage switch opening and closing speed detection to detect the mechanical motion characteristics of the opening and closing action mechanism for the data collected by laser speed measurement, and analyzes and judges the performance changes of the opening and closing action mechanism. First, the speed data of the opening and closing of the high-voltage switch is collected contactlessly by the laser sensor. However, in reality, since it takes a long time to detect the high-voltage circuit breaker, and the detection also involves the conversion between different data, the storage requirements for the data are extremely high, which greatly increases the hardware investment. And with the continuous increase of data, it is also extremely difficult to maintain, which makes the practical application of non-contact opening and closing speed detection of high-voltage switches have certain resistance. Therefore, it is urgent to design a new data training technology based on the opening and closing speed detection of high-voltage switches, train the opening and closing speed data unit data of the high-voltage switch collected by laser sensing, improve the detection accuracy, and judge the performance change trend of the action mechanism. Summary of the invention

[0004] To solve the above problems, the present invention provides a data training method based on the detection of the opening and closing speeds of high-voltage switches, that is, a data storage and analysis method for laser reflection and diffuse reflection in the non-contact detection of the opening and closing speeds of high-voltage switches. By forming diffuse reflection of laser light at the high-voltage circuit breaker, and based on the reflected light, the speed of the reflected light and the position changes of the opening and closing of the high-voltage circuit breaker are obtained, so as to realize the non-contact detection of the opening and closing speeds of the high-voltage circuit breaker. The data training for realizing the laser non-contact detection and analysis includes the following steps:

[0005] Step 1, the present invention considers storing the data collected by laser emission and the data collected by laser diffuse reflection to the opening and closing positions of the high-voltage switch in the form of a matrix tensor product;

[0006] Step 2, convert the collected data stored in the form of a matrix tensor product according to Step 1 into data units for calculating the opening and closing speeds of the high-voltage switch;

[0007] Step 3, perform data query and analysis on the data of the storage unit obtained according to Step 2, and give query results and analysis results according to the requirements of the opening and closing speed detection of the high-voltage switch;

[0008] Step 4, submit the query results and analysis results to the host for decision-making;

[0009] Step 5, send the decision result of the host decision to the maintenance personnel to decide whether to repair or replace the operating mechanism of the high-voltage switch.

[0010] Among them, Step 1 specifically includes:

[0011] The data collected by laser emission and the data collected by laser diffuse reflection to the opening and closing positions of the high-voltage switch should be B = <m, t, d>, where m represents the position identifier of the laser reflection point, t is the collected time, and d represents the collected data;

[0012] B1, perform data conversion, and the converted data can be expressed as

[0013] B = <B 1,v ,B 2,v ,…,B K,v >,

[0014] K represents the time membership dimension of data mapping;

[0015] B2, perform a transformation on the membership method of the laser reflection and diffuse reflection positions for each parallel data to obtain:

[0016] B 1,v ={e 1,v,1 ,e 1,v,2 ,…,e 1,v,M},

[0017] Where M represents the geographical affiliation dimension of data mapping;

[0018] B3, calculate the difference of each laser reflection point position, the initial value of the reflection point is

[0019] B 1,v ={e 1,v,1 ,e 1,v,2 ,…,e 1,v,M},

[0020] The minimum value is

[0021] e intial =Min{e 1,v,1 ,e 1,v,2 ,…,e 1,v,M},

[0022] Further calculation to obtain the difference data vector

[0023] ΔB 1,v ={Δe 1,v,1 ,Δe 1,v,2 ,…,Δe 1,v,M};

[0024] B4, builds the final tensor matrix

[0025]

[0026] Wherein, step 2 specifically includes:

[0027] Decompose the calculation and store the matrix cells in the data cell format

[0028]

[0029] Perform low-rank decomposition.

[0030] Wherein, step 3 specifically includes:

[0031] Perform data query and analysis on the storage unit data after low-rank decomposition obtained in step 2, and provide query results and analysis results according to the requirements of high-voltage switch opening and closing speed detection;

[0032] The final query and decomposition results are obtained by low-rank decomposing the data unit tensor matrix collected by laser reflection and diffuse reflection:

[0033] B (m) =D (m) Z (m) +E (m) ,

[0034] in(·) (m) represents a modulo m operation on a multi-element matrix.

[0035]

[0036] in

[0037] are independent data whose elements follow a standard location normal distribution:

[0038]

[0039] Among them, t m It is also a normal distribution with standard position; is the unit matrix. The decomposition result is obtained by comparing the position of the normal mechanical state that the laser should reflect and the speed of its opening and closing action, which is the normal value state of the detection standard, achieved within the unit standard time.

[0040] Wherein, step 4 specifically includes:

[0041] The query results and analysis results are submitted to the host for decision making

[0042] According to the laser reflection and diffuse reflection position decomposition matrix obtained in step 3

[0043] B (m) =D (m) Z (m) +E (m) ,

[0044] Perform eigenvalue decision and get E (m) The eigenvalue of

[0045]

[0046] and D (m) Z (m) The eigenvalue of

[0047]

[0048] Wherein, step 5 specifically includes:

[0049] Based on the host's decision, the data is sent to the operation and maintenance personnel via 4G wireless transmission to decide whether to repair or replace the action mechanism of the high-voltage switch.

[0050] The present invention discloses the following technical effects:

[0051] The data training method based on the opening and closing speed detection of a high-voltage switch designed in the present invention provides a new method for the data storage and analysis training of the opening and closing speed detection technology of the mechanical performance of the high-voltage switch action mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1 It is a schematic diagram of the method flow of the present invention.

[0054] Figure 2 It is a schematic diagram of the method analysis training calculation steps described in the present invention. DETAILED DESCRIPTION

[0055] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0056] like Figure 1 As shown, the present invention provides a data training method based on the detection of the opening and closing speed of a high-voltage switch, wherein a laser forms diffuse reflection at the high-voltage circuit breaker, and the reflected light speed and the position change of the opening and closing of the high-voltage circuit breaker are obtained according to the reflected light, so as to realize the non-contact detection of the opening and closing speed of the high-voltage circuit breaker. The data training for realizing the laser non-contact detection analysis includes the following steps:

[0057] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A data training method based on high-voltage switch opening and closing speed detection, characterized in that: By measuring the position and speed data of the high-voltage switch opening and closing, and the position and speed data of the reflected light after diffuse reflection, the data is converted into a storage data unit in the form of a matrix tensor product. The data unit data is queried and analyzed through low-rank decomposition. According to the requirements of the high-voltage switch opening and closing speed detection, the query results and analysis results are given. The query results and analysis results are submitted to the host for decision-making. According to the decision of the host, the action mechanism of the high-voltage switch is repaired or replaced to realize the training of the high-voltage switch opening and closing speed detection data. The matrix tensor product storage unit data training for the high-voltage switch opening and closing speed detection includes the position and speed data related to the laser, and the position and speed data of the reflected light after diffuse reflection.

2. A data training method based on high-voltage switch opening and closing speed detection includes the following steps: Step 1, the present invention considers the storage method of the laser emission collection data and the laser diffuse reflection to the high-voltage switch opening and closing position collection data. In this step, the data is collected at different positions to form each independent data storage unit, and is stored in the form of matrix tensor product; Step 2, converting the collected data stored in the form of matrix tensor product in step 1 into a data unit for calculating the opening and closing speed of the high-voltage switch; Step 3, performing data query and analysis based on the storage unit data obtained in step 2, and providing query results and analysis results according to the requirements of high-voltage switch opening and closing speed detection; Step 4: Submit the query results and analysis results to the host for decision making; Step 5: The decision result of the host decision is detected based on the mechanical characteristics of the high-voltage switch, displayed on the interface, and sent to the operation and maintenance personnel in real time through 4G wireless communication. The operation and maintenance personnel decide to repair or replace the action mechanism of the high-voltage switch based on the warning information of the decision.

3. The data training method based on high-voltage switch opening and closing speed detection according to claim 1 is characterized in that: in, Step 1 specifically includes: The data collected from the moment the laser sensor emits the laser and the data collected when the laser is diffusely reflected to the opening and closing positions of the high-voltage switch should be B=<m,t,d> Where m represents the position mark of the laser reflection point, t is the acquisition time when the sensor starts to detect, d represents the collected reflection and diffuse reflection data, and the data conversion process for establishing the tensor matrix is ​​carried out according to B1, B2, B3, and B4; B1, firstly, the data stored in the data unit is transformed, wherein the transformed data can be expressed as, where K represents the time belonging dimension of the transformed data mapping: B=<B 1,v ,B 2,v ,…,B K,v > B2, further transform the position affiliation of each parallel data to obtain B 1,v ={and 1,v,1 ,And 1,v,2 ,…,And 1,v,M } Where M represents the geographical affiliation dimension of data mapping; B3, further calculate the difference of each reflection point, the initial value is B 1,v ={and 1,v,1 ,And 1,v,2 ,…,And 1,v,M } The minimum value is of intial =Min{e 1,v,1 ,of 1,v,2 ,…,of 1,v,M } Get the difference data vector ΔB 1,v ={Δe 1,v,1 ,Δe 1,v,2 ,…,Δe 1,v,M }; B4, further established the final tensor matrix as: The conversion of data unit data is completed by calculation, and the data unit data is stored in the form of matrix tensor product; 4. The data training method based on high-voltage switch opening and closing speed detection according to claim 1 is characterized in that: in, Step 2 specifically includes: Low rank factorization tensor matrix 5. The data training method based on high-voltage switch opening and closing speed detection according to claim 1 is characterized in that: in, Step 3 specifically includes: Perform data query and analysis based on the storage unit data obtained in step 2, and provide query results and analysis results according to the requirements of high-voltage switch opening and closing speed detection; After low-rank decomposition of the tensor matrix, the final query and decomposition result are obtained as follows: B (m) =D (m) Z (m) +E (m) in(·) (m) Represents a modulo m operation on a multidimensional matrix, resulting in: in are independent data, and the elements in them follow the standard normal distribution: Among them, t m It also follows the standard normal distribution; is the identity matrix.

6. The data training method based on high-voltage switch opening and closing speed detection according to claim 1 is characterized in that: in, Step 4 specifically includes: The query results and analysis results are submitted to the high-voltage switch mechanical characteristics detection host for decision-making. The address decomposition matrix obtained according to step 3 is: B (m) =D (m) Z (m) +E (m) Further eigenvalue decision is performed to obtain E (m) The eigenvalue of and D (m) Z (m) The eigenvalue of 7. The data training method based on high-voltage switch opening and closing speed detection according to claim 1 is characterized in that: in, Step 5 specifically includes: Based on the host's decision, the data is sent to the operation and maintenance personnel via 4G wireless transmission to decide whether to repair or replace the action mechanism of the high-voltage switch.

Images