Intelligent enclosed fuse and residual life prediction method

Abstract

The invention discloses an intelligent closed fuse. The fuse comprises a hollow insulating outer sleeve, a fuse body and a fuse body, the fusing device is arranged in the insulating jacket in a sleeving manner; the acquisition device is used for acquiring current information and temperature information of the fusing device; and two ends of the fusing device are connected with leads, namely an upper connecting wire at the front end and a lower connecting wire at the rear end. Compared with an open fuse, the fuse has better environmental adaptability, the real-time state of the fuse carrier can be acquired through the acquisition device, so that the detection and maintenance efficiency is improved, and furthermore, the invention further provides a residual life prediction method, so that the state of the fuse carrier can be dynamically acquired according to the state information acquired by the acquisition device in real time, and the detection and maintenance efficiency is improved. And the effects of finding and overhauling in advance are achieved. In addition, the invention further provides a closing operation rod which is used for fixing the fuse carrier more conveniently.

Description

Technical Field

[0001] The present invention belongs to the technical field of fuses, and in particular relates to an intelligent enclosed fuse and a remaining life prediction method. Background Art

[0002] 10kV enclosed fuse is suitable for outdoor overhead power distribution system with AC 50Hz and rated voltage of 10kV, for overload and short circuit protection of lines and power transformers. When short circuit or current is too large, the fault current causes the fuse to melt quickly to form an arc. The polymer gas-generating material built into the fuse tube instantly releases a large amount of gas under the high temperature of the arc, and blows strongly longitudinally along the pipeline, and the arc is quickly elongated and extinguished. After the fuse is melted, the lower contact of the fuse carrier loses its pulling force, and the fuse carrier falls to form a fracture. When it is necessary to pull the load, the fuse carrier is pulled apart with an insulating rod. At this time, the upper contact of the fuse carrier is separated from the upper contact group of the base, and an arc is generated between the two. The arc is elongated in the slit formed by the arc extinguishing tube and the arc extinguishing rod of the base. At the same time, the arc extinguishing tube and the arc extinguishing rod generate gas, and the arc is extinguished when the current passes through zero. After normal disconnection or operation of the product, the fuse carrier pops out to form a fracture, which plays a role in safety protection for the maintenance of lines and equipment. The enclosed fuse is mainly composed of a silicone rubber composite jacket base and a fuse carrier. The fuse carrier is placed inside the base and has good electrical insulation, safety and weather resistance.

[0003] In the existing 10kV distribution line system, the distribution transformer substation basically uses drop-out fuses. Due to the exposed contacts, drop-out fuses are suitable for environments with basically no interference in the surrounding area (such as branches, small animals, etc.), and have high requirements for air quality. They are suitable for dangerous environments without conductive dust, sand, corrosive gases, flammable and explosive environments. They are not suitable for coastal high salt fog and some severely polluted salt fog areas. Drop-out fuses also have the risk of loosening of porcelain insulator hardware and fracture of porcelain parts. During normal use of the product, due to the difference in thermal expansion coefficients, porcelain parts and hardware will become loose over time, and in severe cases, the insulator will detach. In addition, there may be microcracks in the porcelain parts themselves during the molding and firing process, causing stress concentration and the possibility of fracture.

[0004] The enclosed fuse overcomes these shortcomings of the traditional drop-out fuse and eliminates the influence of environmental conditions on the performance of the fuse. However, due to its enclosed structure, its overall status cannot be inspected during inspection. Usually, when inspecting the drop-out fuse, it is mainly to check whether the base is installed loosely, whether the porcelain parts have cracks, flashovers, damage and dirt; whether the connection points are good; whether the contact between the contacts is good, whether there is overheating, burning, melting; whether the fuse carrier is abnormal, whether the fuse tube is cracked, delaminated, aged, deformed, etc. For enclosed fuses, due to their enclosed structure, it is impossible to inspect whether the contact between the contacts is good, whether there is overheating, burning, melting and the status of the fuse carrier. Therefore, for enclosed fuses, it is a very urgent need to realize operating status and life monitoring.

[0005] In addition, the distribution network fault location and load monitoring system is an important model for the construction of distribution network automation. It can play the basic functions of distribution production management, including rapid location and search of line faults, and line operation load monitoring. As an important part of the distribution automation function, distribution network fault search and location are the main ways to improve the efficiency of distribution network fault repair and the level of distribution production management. Traditional fault repair mainly learns that there is a fault in the line through dispatch or user report, and then spends a lot of manpower and material resources to conduct a dragnet inspection of the line or find the fault point step by step, and then restore power supply after the fault is handled. This fault handling method makes the fault repair work relatively passive, and it takes a lot of time to determine the fault point, which is time-consuming and laborious.

[0006] The intelligent monitoring system of fuses is based on the principle of "advanced, stable, reliable, practical and economical", with the goal of improving power supply reliability, power supply quality, management service level and economic benefits of enterprises. It strives to change the passive situation of "hasty response and response with unclear purpose" in the previous distribution network fault repair work, so as to achieve the accuracy and speed of distribution network fault repair work and improve the efficiency of distribution network repair work. The background monitoring software can realize remote fault automatic positioning to meet the needs of distribution network equipment operation status monitoring.

[0007] The outer insulation layer of the composite sheath fuse is made of injection-molded outdoor silicone rubber material with excellent weather resistance, which reduces the risk of the product exploding when it is opened; the product is fully insulated as a whole, and the contact parts are all placed inside the sheath, which improves the insulation degree of the distribution network, can effectively reduce the installation spacing of equipment, save space, and reduce the accident rate; the upper and lower terminals can be designed as a lead structure, which is an integrated design structure with the fuse body, with significantly improved anti-pollution performance and excellent weather resistance. Summary of the invention

[0008] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent enclosed fuse, characterized in that the fuse comprises:

[0009] Hollow insulating jacket;

[0010] A fuse device, sleeved in the insulating jacket;

[0011] A collection device for collecting current information and temperature information of the fuse device;

[0012] Lead wires are connected to two ends of the fuse device, namely an upper wire at the front end and a lower wire at the rear end.

[0013] Furthermore, the collection device includes a collection device sleeved on the upper lead wire, and a sheath wrapped and fixed on the collection device, a sealing portion is provided between the sheath and the lead wire, the collection device and the insulating jacket, and the sealing portion is used to seal the gap between the sheath and the lead wire, the collection device and the insulating jacket; the collection device is arranged between the upper lead wire and the insulating jacket or between the lower lead wire and the insulating jacket.

[0014] Furthermore, the fuse device includes an upper contact seat connected to the upper wire and a lower contact seat connected to the lower wire, and also includes a fuse carrier arranged between the upper contact seat and the lower contact seat, one end of the fuse carrier is connected to the upper contact seat through an upper contact, and the other end is connected to the lower contact seat through a lower contact.

[0015] Furthermore, the insulating jacket is made of vulcanized silicone rubber; the lead wire is an insulated wire; the lead wire is a copper rod; and a mounting frame is provided on the outer surface of the insulating jacket.

[0016] Furthermore, the intelligent enclosed fuse is also provided with a closing operation rod, and the closing operation rod includes:

[0017] An adjusting nut is embedded with a plurality of balls and springs, the balls are pressed by the springs, and the other ends of the balls are pressed against the fuse carrier nut of the fuse carrier to fix the fuse carrier;

[0018] An operating rod, one end of which is fixedly connected to one end of the adjusting nut;

[0019] A connecting member, one end of which is fixedly connected to the other end of the operating rod;

[0020] An extension rod has one end fixedly connected to the other end of the connecting member.

[0021] A method for predicting the remaining life of an intelligent enclosed fuse, the intelligent enclosed fuse described above, is characterized by comprising the following steps:

[0022] Step S1: Obtain historical current information and temperature information passing through the fuse carrier through a collection device, and mark the corresponding remaining life;

[0023] Step S2: importing the acquired current information and temperature information for preprocessing to obtain remaining life characteristic information;

[0024] Step S3: Sequence segmentation is performed on the remaining life feature information to obtain subsequence features;

[0025] Step S4: importing subsequence features into the gated convolution unit to obtain the extracted feature vector;

[0026] Step S5: importing the extracted feature vector in step S4 into the position editor of the fault diagnosis model to inject position information and obtain an injected feature vector;

[0027] Step S6: injecting the injected feature vector in step S5 into the transformer layer of the fault diagnosis model for prediction, and obtaining the prediction result, that is, the remaining life of the fuse carrier;

[0028] Step S7: construct a loss function, and minimize the loss function to optimize the parameters of the model;

[0029] Step S8: importing the real-time current information and temperature information collected by the collection module to obtain the predicted remaining life of the fuse carrier.

[0030] Furthermore, step S3 is specifically as follows:

[0031] The remaining life feature information is decomposed into time series by PatchTST, which is expressed as:

[0032] X i =X t：t+L ;

[0033]

[0034] Among them, X i represents the i-th subsequence extracted from the time series X, time series X∈R TxF , T represents the time step, F represents the number of features, t represents the time step corresponding to the current sequence, R represents the shape or dimension of the input time series data, X t：t+L represents a subsequence of length L, and N is the total number of subsequences.

[0035] Furthermore, step S4 is specifically as follows:

[0036] Step S41: Perform a one-dimensional convolution on the acquired subsequence, expressed as:

[0037] h i =1DCNN(X i )

[0038] Among them, h irepresents the convolution feature vector, and 1DCNN represents the one-dimensional convolution layer;

[0039] Step S42: The convolution feature vector processed by the one-dimensional convolution layer is imported into the gate control unit and passes through the reset gate and update gate for controlling the information flow. The formulas of the reset gate and update gate are as follows:

[0040] r i =σ(W r ·h i +V r ·x i +b r );

[0041] u i =σ(W u ·h i +V u ·x i +b u );

[0042] Among them, r i represents the reset gate, u i represents the update gate, σ represents the sigmoid activation function, W r , W u , V r , V u Represent different weight matrices, b r , b u Represent different bias terms respectively;

[0043] Step S43: Perform residual connection, expressed as:

[0044] hh i =h i ⊙u i +x i ⊙r i ;

[0045] hh fin =W e ·hh i +b e ;

[0046] Among them, hh i Indicates the initial extraction of feature vectors, hh fin represents the intermediate extracted feature vector, W e represents the residual weight matrix, ⊙ represents the element-by-element product, b e Represents the bias of the residual connection;

[0047] Step S44: Padded processing is performed on the intermediate extracted feature vector of step S33, which is expressed as:

[0048]

[0049] in, represents the extracted feature vector, Represents x after filling i .

[0050] Furthermore, step S5 is specifically as follows:

[0051] The position encoder generates a unique encoding of the position and dimension of the subsequence through a combination of sine and cosine functions. The position encoding PE is formalized as:

[0052] When i is an even number:

[0053]

[0054] When i is an odd number:

[0055]

[0056] Among them, PE(pos, 2i) represents the even position encoding, PE(pos, 2i+1) represents the odd position encoding, and d model represents the dimension of the position encoder, pos represents the position of the subsequence, and i represents the dimension of the subsequence.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] The present invention provides a closed fuse, which has better environmental adaptability than an open fuse, and can obtain the real-time status of the fuse carrier through a collection device, thereby improving the detection and maintenance efficiency. Furthermore, the present invention also provides a remaining life prediction method, so that the status of the fuse carrier can be dynamically obtained according to the status information obtained in real time by the collection device, which plays the role of early discovery and maintenance. In addition, the present invention also provides a closing operating rod for more conveniently fixing the fuse carrier. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the overall structure of the enclosed fuse of the present invention.

[0060] Figure 2 The figure is a schematic diagram of the structure of the closing operating rod of the enclosed fuse of the present invention.

[0061] Figure 3 It is a schematic structural diagram of the closing operating rod of the enclosed fuse of the present invention from another perspective. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] Reference Figure 1 , an intelligent enclosed fuse, in one example, the fuse comprises:

[0064] A hollow insulating jacket 1;

[0065] A fuse device, sleeved in the insulating jacket 1;

[0066] A collection device 2, for collecting current information and temperature information of the fuse device;

[0067] Lead wires are connected to two ends of the fuse device, namely an upper wire 6 at the front end and a lower wire 7 at the rear end.

[0068] Furthermore, in one example, the collection device 2 includes a collection device 2 sleeved on the upper lead 6, and a sheath wrapped and fixed on the collection device 2, a sealing portion is provided between the sheath and the lead, the collection device 2 and the insulating jacket 1, and the sealing portion is used to seal the gap between the sheath and the lead, the collection device 2 and the insulating jacket 1; the collection device 2 is arranged between the upper lead and the insulating jacket 1 or between the lower lead and the insulating jacket 1.

[0069] Furthermore, in one example, the fuse device includes an upper contact seat 4 connected to an upper wire 6 and a lower contact seat 5 connected to a lower wire 7, and also includes a fuse carrier 3 arranged between the upper contact seat 4 and the lower contact seat 5, one end of the fuse carrier 3 is connected to the upper contact seat 4 through an upper contact, and the other end is connected to the lower contact seat 5 through a lower contact.

[0070] Furthermore, in one example, the insulating jacket 1 is made of vulcanized silicone rubber; the lead wire is an insulated wire; the lead wire is a copper rod; and a mounting frame 8 is provided on the outer surface of the insulating jacket 1.

[0071] For further reference, Figure 2 and Figure 3 In one example, the intelligent enclosed fuse is further provided with a closing operation rod, and the closing operation rod includes:

[0072] The adjusting nut 9 has a plurality of balls 902 and a spring 901 embedded therein, and the spring 901 applies pressure to the balls 902, and the other end of the balls 902 abuts against the nut of the fuse carrier 3 to fix the fuse carrier 3;

[0073] An operating rod 10, one end of which is fixedly connected to one end of the adjusting nut 9;

[0074] A connecting member 11, one end of which is fixedly connected to the other end of the operating rod 10;

[0075] One end of the extension rod 12 is fixedly connected to the other end of the connecting member 11 .

[0076] A method for predicting the remaining life of an intelligent enclosed fuse, in one example, comprises the following steps:

[0077] Step S1: acquiring current information and temperature information passing through the fuse through the acquisition device 2;

[0078] Step S2: importing the acquired current information and temperature information for preprocessing to obtain remaining life characteristic information;

[0079] Step S3: Sequence segmentation is performed on the remaining life feature information to obtain subsequence features;

[0080] Step S4: importing subsequence features into the gated convolution unit to obtain the extracted feature vector;

[0081] Step S5: importing the extracted feature vector in step S4 into the position editor of the fault diagnosis model to inject position information and obtain an injected feature vector;

[0082] Step S6: injecting the injected feature vector in step S5 into the transformer layer of the fault diagnosis model for prediction, and obtaining the prediction result, that is, the remaining life of the fuse;

[0083] Step S7: Construct a loss function and minimize the loss function to optimize the parameters of the model.

[0084] Furthermore, in one example, step S3 is specifically as follows:

[0085] The remaining life feature information is decomposed into time series by PatchTST, which is expressed as:

[0086] X i =X t：t+L ;

[0087]

[0088] Among them, X i represents the i-th subsequence extracted from the time series X, time series X∈R TxF , T represents the time step, F represents the number of features, t represents the time step corresponding to the current sequence, R represents the shape or dimension of the input time series data, X t：t+L represents a subsequence of length L, and N is the total number of subsequences.

[0089] Furthermore, in one example, step S4 is specifically as follows:

[0090] Step S41: Perform a one-dimensional convolution on the acquired subsequence, expressed as:

[0091] h i =1DCNN(X i )

[0092] Among them, h i represents the convolution feature vector, and 1DCNN represents the one-dimensional convolution layer;

[0093] Step S42: The convolution feature vector processed by the one-dimensional convolution layer is imported into the gate control unit and passes through the reset gate and update gate for controlling the information flow. The formulas of the reset gate and update gate are as follows:

[0094] r i =σ(W r @h i +V r ·x i +b r );

[0095] u i =σ(W u ·h i +V u ·x i +b u );

[0096] Among them, r i represents the reset gate, u i represents the update gate, σ represents the sigmoid activation function, W r , W u , V r , V u Represent different weight matrices, b r , b u Represent different bias terms respectively;

[0097] Step S43: Perform residual connection, expressed as:

[0098] hh i =h i ⊙u i +x i ⊙r i ;

[0099] hh fin =W e ·hh i +b e ;

[0100] Among them, hh i Indicates the initial extraction of feature vectors, hh fin represents the intermediate extracted feature vector, W e represents the residual weight matrix, ⊙ represents the element-by-element product, b e Represents the bias of the residual connection;

[0101] Step S44: Padded processing is performed on the intermediate extracted feature vector of step S33, which is expressed as:

[0102]

[0103] in, represents the extracted feature vector, Represents x after filling i .

[0104] Furthermore, in one example, step S5 is specifically as follows:

[0105] The position encoder generates a unique encoding of the position and dimension of the subsequence through a combination of sine and cosine functions. The position encoding PE is formalized as:

[0106] When i is an even number:

[0107]

[0108] When i is an odd number:

[0109]

[0110] Among them, PE(pos, 2i) represents the even position encoding, PE(pos, 2i+1) represents the odd position encoding, and d model represents the dimension of the position encoder, pos represents the position of the subsequence, and i represents the dimension of the subsequence.

[0111] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent enclosed fuse, characterized in that: The fuse comprises: Hollow insulating jacket; A fuse device, sleeved in the insulating jacket; A collection device for collecting current information and temperature information of the fuse device; Lead wires are connected to two ends of the fuse device, namely an upper wire at the front end and a lower wire at the rear end.

2. The intelligent enclosed fuse according to claim 1, characterized in that: The collection device includes a collection device sleeved on the upper lead wire, and a sheath wrapped and fixed on the collection device. A sealing portion is provided between the sheath and the lead wire, the collection device and the insulating jacket. The sealing portion is used to seal the gap between the sheath and the lead wire, the collection device and the insulating jacket. The collection device is arranged between the upper lead wire and the insulating jacket or between the lower lead wire and the insulating jacket.

3. The intelligent enclosed fuse according to claim 2, characterized in that: The fuse device includes an upper contact seat connected to the upper wire and a lower contact seat connected to the lower wire, and also includes a fuse carrier arranged between the upper contact seat and the lower contact seat, one end of the fuse carrier is connected to the upper contact seat through an upper contact, and the other end is connected to the lower contact seat through a lower contact.

4. The intelligent enclosed fuse according to claim 1, characterized in that: The insulating jacket is made of vulcanized silicone rubber; the lead wire is an insulated wire; the lead wire is a copper rod; and a mounting frame is arranged on the outer surface of the insulating jacket.

5. The intelligent enclosed fuse according to claim 1, characterized in that: The intelligent enclosed fuse is also provided with a closing operating rod, which includes: An adjusting nut is embedded with a plurality of balls and springs, the balls are pressed by the springs, and the other ends of the balls are pressed against the fuse carrier nut of the fuse carrier to fix the fuse carrier; An operating rod, one end of which is fixedly connected to one end of the adjusting nut; A connecting member, one end of which is fixedly connected to the other end of the operating rod; An extension rod has one end fixedly connected to the other end of the connecting member.

6. A method for predicting the remaining life of an intelligent enclosed fuse, used for an intelligent enclosed fuse as claimed in any one of claims 1 to 5, characterized in that: The following steps are involved: Step S1: obtaining current information and temperature information passing through the fuse through a collection device; Step S2: importing the acquired current information and temperature information for preprocessing to obtain remaining life characteristic information; Step S3: Sequence segmentation is performed on the remaining life feature information to obtain subsequence features; Step S4: importing subsequence features into the gated convolution unit to obtain the extracted feature vector; Step S5: importing the extracted feature vector in step S4 into the position editor of the fault diagnosis model to inject position information and obtain an injected feature vector; Step S6: injecting the injected feature vector in step S5 into the transformer layer of the fault diagnosis model for prediction to obtain a prediction result, i.e., the remaining life of the fuse; Step S7: Construct a loss function and minimize the loss function to optimize the parameters of the model.

7. The method for predicting the remaining life of an intelligent enclosed fuse according to claim 6, characterized in that: Step S3 is specifically as follows: The remaining life feature information is decomposed into time series by PatchTST, which is expressed as: X i =X t：t+L ; Among them, X i represents the i-th subsequence extracted from the time series X, time series X∈R TxF , T represents the time step, F represents the number of features, t represents the time step corresponding to the current sequence, R represents the shape or dimension of the input time series data, X t：t+L represents a subsequence of length L, and N is the total number of subsequences.

8. The method for predicting the remaining life of an intelligent enclosed fuse according to claim 7, characterized in that: Step S4 is specifically as follows: Step S41: Perform a one-dimensional convolution on the acquired subsequence, expressed as: h i =1DCNN(X i ) Among them, h i represents the convolution feature vector, and 1DCNN represents the one-dimensional convolution layer; Step S42: The convolution feature vector processed by the one-dimensional convolution layer is imported into the gate control unit and passes through the reset gate and update gate for controlling the information flow. The formulas of the reset gate and update gate are as follows: r i =σ(W r ·h i +V r ·x i +b r ); u i =σ(W u ·h i +V u ·x i +b u ); Among them, r i represents the reset gate, u i represents the update gate, σ represents the sigmoid activation function, W r , W u , V r , V u Represent different weight matrices, b r , b u Represent different bias terms respectively; Step S43: Perform residual connection, expressed as: hh i =h i ⊙u i +x i ⊙r i ; hh fin =W e ·hh i +b e ; Among them, hh i represents the initial extraction of feature vectors, hh fin represents the intermediate extracted feature vector, W e represents the residual weight matrix, ⊙ represents the element-by-element product, b e Represents the bias of the residual connection; Step S44: Padded processing is performed on the intermediate extracted feature vector of step S33, which is expressed as: in, represents the extracted feature vector, Represents x after filling i .

9. The method for predicting the remaining life of an intelligent enclosed fuse according to claim 8, characterized in that: Step S5 is specifically as follows: The position encoder generates a unique encoding of the position and dimension of the subsequence through a combination of sine and cosine functions. The position encoding PE is formalized as: When i is an even number: When i is an odd number: Among them, PE(pos, 2i) represents the even position encoding, PE(pos, 2i+1) represents the odd position encoding, and d model represents the dimension of the position encoder, pos represents the position of the subsequence, and i represents the dimension of the subsequence.

Images