Relay predictive maintenance system and method based on contact wear

Abstract

The invention relates to a relay predictive maintenance system and method based on contact wear, and the system comprises a contact wear parameter collection module, a counter-force calculation module, a counter-force step identification module, a contact wear value calculation module, a data transmission module, and a contact life prediction module. Each relay is connected with an independent contact service life prediction module, the counter-force calculation module, the counter-force step identification module and the contact wear value calculation module are arranged at the edge end, and the contact service life prediction modules are arranged on the cloud platform; the contact wear parameter acquisition module is used for acquiring circuit parameters when each relay is switched on; the counter-force calculation module is used for rewriting a counter-force formula related to contact abrasion on the basis of an Alembert mechanical motion equation and calculating the counter-force borne by a movable iron core in the closing process of each relay; and the counter-force step identification module identifies a step point of the counter-force based on a least square method, namely a collision time point of the contact in the relay.

Description

Technical Field

[0001] The present invention relates to the technical field of switch electrical appliances, and in particular to a relay predictive maintenance system and method based on contact wear. Background Art

[0002] The power grid system is becoming more intelligent and complex, and the relay is one of its core components. Its performance indicators directly affect the safety and stability of the entire power grid control system. During the operation of the relay, the arc of opening and closing will continuously corrode the contacts, causing the end of its electrical life and leading to failure. In response to this problem, a large number of scholars have done a lot of research on the design of the body and the prediction of electrical life.

[0003] Regarding contact wear observation, the more commonly used method is to obtain the direct relationship between the operating parameters such as the pull-in time, arcing time, arcing energy, and contact resistance and the contact use time through experiments, and use feature extraction and artificial intelligence algorithms to indirectly observe the contact wear of the relay. This type of method builds a prediction algorithm from the direct mapping relationship between the remaining use time and the operating characteristics, without prior knowledge, but the training process requires a large amount of data.

[0004] Regarding the prediction of electrical life, currently commonly used methods include statistical regression and machine learning algorithms, the core idea of ​​which is to predict the remaining service life of the current contactor based on the time series data of the historical operating status of the relay. Currently commonly used methods include long short-term memory neural network, recurrent neural network, Winner process and Bayes process, etc. However, these prediction methods use the same model for relays in different operating environments, resulting in poor model prediction results for states under different operating conditions.

[0005] Therefore, this patent designs a relay predictive maintenance technology based on contact wear, which combines contact wear observation, Internet of Things technology and life prediction technology, and specifies unique observation models for relays under different working conditions, which plays a vital role in the reliable operation of relays. Summary of the invention

[0006] In order to solve the above problems existing in the prior art, the present invention proposes a relay predictive maintenance system and method based on contact wear.

[0007] The technical solution of the present invention is as follows:

[0008] On the one hand, the present invention proposes a relay predictive maintenance system based on contact wear, characterized in that it includes a contact wear parameter acquisition module, a reaction force calculation module, a reaction force step identification module, a contact wear value calculation module, a data transmission module and a contact life prediction module, wherein each relay is connected to a separate contact life prediction module, the reaction force calculation module, the reaction force step identification module and the contact wear value calculation module are deployed at the edge end, and the contact life prediction module is arranged on a cloud platform;

[0009] Contact wear parameter collection module, collects circuit parameters when each relay is closed;

[0010] The reaction force calculation module rewrites the reaction force formula related to contact wear based on the D'Alembert mechanical motion equation and calculates the reaction force on the moving iron core during the closing process of each relay;

[0011] The reaction force step identification module identifies the step point of the reaction force, i.e. the collision moment of the contacts in the relay, based on the least square method;

[0012] The contact wear value calculation module calculates the wear value of the corresponding contact by using the reaction force on the moving iron core of each relay and the step point of the reaction force;

[0013] The data transmission module builds a wide-area Internet of Things based on Lora communication and 4G communication, and transmits the contact wear values ​​collected by all contactors to the cloud platform;

[0014] The contact life prediction module predicts the remaining service life of the corresponding relay contact based on the historical wear value of each relay.

[0015] As a preferred implementation, the reaction force formula related to contact wear is rewritten based on the D'Alembert mechanical motion equation as follows:

[0016]

[0017] In the formula, x is the displacement distance of the moving iron core, t is the displacement time, F X is the electromagnetic attraction, F f For the reaction force.

[0018] As a preferred implementation, the step of identifying the step point of the reaction force based on the least square method, that is, the collision moment of the contacts in the relay, is specifically as follows:

[0019] Assume a point as the step point of the original reaction force-displacement discrete data, generate a basis function containing the same step point, use the basis function to achieve the least squares optimal fitting of the original reaction force-displacement discrete data, and record the fitting error; use the three-point search method to find the point in the original reaction force-displacement discrete data where the fitting error of the original reaction force-displacement discrete data using the basis function is the smallest, and this point is the step point of the original reaction force-displacement discrete data;

[0020] The formula of the basis function is as follows:

[0021]

[0022] The constraints are:

[0023] x0≤w slip ≤x n ,z slip ≠z zlip1

[0024] In the formula, w is the function variable, w slip is the function segmentation point, z slip is the function step left value, z slip1 is the right value of the function step; p1~p6 are coefficients, x0, x n is the distribution range of the original reaction force-displacement discrete data.

[0025] As a preferred implementation, the step of calculating the contact wear value using the reaction force of the moving iron core and the reaction force step point is specifically as follows:

[0026] The reaction force-displacement curve composed of the reaction force-displacement discrete data is used to obtain the closing moment of the contact in combination with the step point of the reaction force-displacement discrete data. The closing moment of the contact is corresponded to the reaction force-displacement curve to obtain the current contact opening distance. Finally, the current contact opening distance is subtracted from the initial contact opening distance to obtain the current contact wear value.

[0027] As a preferred implementation, the step of predicting the remaining service life of the corresponding relay contact based on the historical wear value of each relay is specifically as follows:

[0028] A long short-term memory neural network is used as the prediction model. The wear value and wear time of each contact are combined to train the prediction model. The trained prediction model is used to predict the remaining service life of the contact.

[0029] On the other hand, the present invention proposes a relay predictive maintenance method based on contact wear, the specific steps comprising:

[0030] Collect circuit parameters when each relay is closed;

[0031] Based on the D'Alembert mechanical motion equation, the reaction force formula related to contact wear is rewritten, and the reaction force on the moving iron core during the closing process in each relay is calculated;

[0032] Identify the step point of the reaction force, i.e. the collision moment of the contacts in the relay, based on the least squares method;

[0033] The wear value of the corresponding contact is calculated using the reaction force on the moving iron core of each relay and the step point of the reaction force;

[0034] Build a wide-area Internet of Things based on Lora communication and 4G communication, and transmit the contact wear values ​​collected by all contactors to the cloud platform;

[0035] The remaining service life of the corresponding relay contacts is predicted based on the historical wear value of each relay.

[0036] As a preferred implementation, the reaction force formula related to contact wear is rewritten based on the D'Alembert mechanical motion equation as follows:

[0037]

[0038] In the formula, x is the displacement distance of the moving iron core, t is the displacement time, F X is the electromagnetic attraction, F f For the reaction force.

[0039] As a preferred implementation, the step of identifying the step point of the reaction force based on the least square method, that is, the collision moment of the contacts in the relay, is specifically as follows:

[0040] Assume a point as the step point of the original reaction force-displacement discrete data, generate a basis function containing the same step point, use the basis function to achieve the least squares optimal fitting of the original reaction force-displacement discrete data, and record the fitting error; use the three-point search method to find the point in the original reaction force-displacement discrete data where the fitting error of the original reaction force-displacement discrete data using the basis function is the smallest, and this point is the step point of the original reaction force-displacement discrete data;

[0041] The formula of the basis function is as follows:

[0042]

[0043] The constraints are:

[0044] x0≤w slip ≤x n ,z slip ≠z zlip1

[0045] In the formula, w is the function variable, w slip is the function segmentation point, z slip is the function step left value, z slip1 is the right value of the function step; p1~p6 are coefficients, x0, x n is the distribution range of the original reaction force-displacement discrete data.

[0046] As a preferred implementation, the step of calculating the wear value of the contact using the reaction force of the moving iron core and the reaction force step point is specifically as follows:

[0047] The reaction force-displacement curve composed of the reaction force-displacement discrete data is used to obtain the closing moment of the contact in combination with the step point of the reaction force-displacement discrete data. The closing moment of the contact is corresponded to the reaction force-displacement curve to obtain the current contact opening distance. Finally, the current contact opening distance is subtracted from the initial contact opening distance to obtain the current contact wear value.

[0048] As a preferred implementation, the step of predicting the remaining service life of the corresponding relay contact based on the historical wear value of each relay is specifically as follows:

[0049] A long short-term memory neural network is used as the prediction model. The wear value and wear time of each contact are combined to train the prediction model. The trained prediction model is used to predict the remaining service life of the contact.

[0050] The present invention has the following beneficial effects:

[0051] 1. The present invention quickly calculates the wear value of the contact based on the step point of the reaction force by fitting the curve of the displacement distance of the moving iron core in the relay and the reaction force it receives.

[0052] 2. The present invention uses the least square method to quickly identify the step point of the reaction force, thereby providing a guarantee for the rapid determination of the contact wear value and the stable and safe operation of the relay.

[0053] 3. The present invention uses a long short-term memory neural network and, based on the historical data of the contact wear values ​​of each relay, makes an accurate prediction of the contact usage time, thereby ensuring the stability of the operation of each relay.

[0054] 4. The present invention quickly collects the operating data on each relay through the contactor, and transmits the operating data to the cloud platform in real time through Lora communication and 4G communication for rapid calculation, thereby ensuring the real-time nature of the evaluation results.

[0055] 5. The present invention adopts separate long and short-term memory neural network to construct models for relays under different working conditions, thereby ensuring the accuracy of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is a system structure diagram of the present invention;

[0057] Figure 2 It is the system structure topology diagram and prediction model architecture diagram;

[0058] Figure 3 Flowchart of the algorithm for finding the minimum fitting error for the three-point search method;

[0059] Figure 4 Schematic diagram of calculation principle of contact wear value. DETAILED DESCRIPTION

[0060] 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.

[0061] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0062] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0063] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0064] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0065] Embodiment 1:

[0066] See also Figure 1 , a relay predictive maintenance system based on contact wear, comprising a contact wear parameter acquisition module, a reaction force calculation module, a reaction force step identification module, a contact wear value calculation module, a data transmission module and a contact life prediction module, wherein each relay is connected to a separate contact life prediction module, the reaction force calculation module, the reaction force step identification module and the contact wear value calculation module are deployed at the edge end, and the contact life prediction module is arranged on a cloud platform;

[0067] Contact wear parameter collection module, collects circuit parameters when each relay is closed;

[0068] The reaction force calculation module rewrites the reaction force formula related to contact wear based on the D'Alembert mechanical motion equation and calculates the reaction force on the moving iron core during the closing process of each relay;

[0069] The reaction force step identification module identifies the step point of the reaction force, i.e. the collision moment of the contacts in the relay, based on the least square method;

[0070] The contact wear value calculation module calculates the wear value of the corresponding contact by using the reaction force on the moving iron core of each relay and the step point of the reaction force;

[0071] The data transmission module builds a wide-area Internet of Things based on Lora communication and 4G communication, and transmits the contact wear values ​​collected by all contactors to the cloud platform;

[0072] The contact life prediction module predicts the remaining service life of the corresponding relay contact based on the historical wear value of each relay.

[0073] In this embodiment, the system structure topology diagram and prediction model architecture of this solution are as follows: Figure 2 shown.

[0074] As a preferred implementation of this embodiment, the reaction force formula related to contact wear rewritten based on the D'Alembert mechanical motion equation is specifically:

[0075]

[0076] In the formula, x is the displacement distance of the moving iron core, t is the displacement time, F X is the electromagnetic attraction, F f For the reaction force.

[0077] In this embodiment, the relay satisfies the D'Alembert mechanical motion equation during movement:

[0078]

[0079] Where, ψ is the magnetic flux; u is the coil voltage; i is the coil current; R is the coil resistance; F x is the electromagnetic attraction; F f is the reaction force; m is the equivalent mass of the movable part; v is the speed of the moving iron core. All these data can be collected by the contactor installed on the relay.

[0080] By rewriting the D'Alembert mechanical motion equation, the reaction force expression can be obtained:

[0081]

[0082] According to the reaction force expression, if we want to find the reaction force F f , then it is necessary to obtain the displacement x and electromagnetic suction force F of the moving iron core during the closing process x , and due to the electromagnetic mechanism coil current i, magnetic circuit flux ψ, moving iron core displacement x, electromagnetic suction force F x Therefore, the BP neural network model can be used to design the moving iron core displacement observer and the electromagnetic suction force observer according to this mapping relationship, and the electromagnetic suction force F can be directly observed according to the coil current i and coil voltage u during the contactor closing process. xand the displacement of the moving iron core x. Further, the reaction force F f , the moving iron core displacement x is used to obtain the reaction force-displacement curve.

[0083] During the contactor closing process, the coil voltage u and coil current i signals of the contactor are collected, and the magnetic flux ψ is indirectly observed with the help of the integral flux observer. The (i, ψ) data is input into the moving iron core displacement observer and the electromagnetic suction force observer to observe the moving iron core displacement x and the electromagnetic suction force F. x , take the second-order differential of the displacement of the moving iron core and multiply it by the mass m of the moving iron core to get the resultant force on the moving iron core during its movement. Since the force on the moving iron core conforms to the superposition theorem, the resultant force F is used minus the electromagnetic attraction F x The reaction force F can be analyzed f .

[0084] As a preferred implementation of this embodiment, the step of identifying the step point of the reaction force based on the least square method, that is, the collision moment of the contacts in the relay, is specifically as follows:

[0085] Assume a point as the step point of the original reaction force-displacement discrete data, generate a basis function containing the same step point, use the basis function to achieve the least squares optimal fitting of the original reaction force-displacement discrete data, and record the fitting error; use the three-point search method to find the point in the original reaction force-displacement discrete data where the fitting error of the original reaction force-displacement discrete data using the basis function is the smallest, and this point is the step point of the original reaction force-displacement discrete data;

[0086] The formula of the basis function is as follows:

[0087]

[0088] The constraints are:

[0089] x0≤w slip ≤x n ,z slip ≠z zlip1

[0090] In the formula, w is the function variable, w slip is the function segmentation point, z slip is the function step left value, z slip1 is the right value of the function step; p1~p6 are coefficients, x0, x n is the distribution range of the original reaction force-displacement discrete data.

[0091] In this embodiment, the reaction force and the corresponding displacement are combined into an original discrete data pair. A point is first assumed to be the step point of the original discrete data, and a basis function containing the same step point is generated. The basis function is used to implement the least squares optimal fit for the original discrete data, and the fitting error is recorded. If there is a point in the original discrete data, and the basis function generated by this point has the smallest fitting error for the original discrete data, this point is the step point of the original discrete data.

[0092] Considering that there are a large number of original discrete data, if the fitting calculation is performed point by point, the algorithm will take a long time; and from the fitting properties, it can be seen that there is always a minimum point for the fitting error and the fitting error is a concave function, so the three-point search method can be used to find the fitting error and reduce the algorithm time.

[0093] The algorithm flow of finding the minimum fitting error using the three-point search method is as follows: Figure 3 As shown in the figure, after each closing, the observed reaction force-displacement discrete data (x j ,y(x j )) Import into the algorithm and set the variable x left =x0,x right =x n And calculate x mid 、x mmid ; Then calculate the approximation error under the action of each variable: Set the segmentation points of the basis function to x left ,x right ,x mid ,x mmid , use each basis function to approximate the original discrete data. When the optimal approximation is reached, calculate the average relative error loss and save the average relative error loss at each segment point. left ,loss right, loss mid ,loss mmid ; Since in (x0,x n ), the approximation error is a concave function, so by judging the loss mid and loss mmid The size of can determine the direction with the smallest loss. min <loss mmid When the smaller value is in loss mind The left side of x right =x mmid , and vice versa; finally judge x left With x right If x left With x right If the distance is far, it proves that there is still a minimum point, so continue to loop to solve; if the distance meets the requirement, the step point x is found. left .

[0094] As a preferred implementation of this embodiment, the step of calculating the wear value of the contact using the reaction force of the moving iron core and the reaction force step point is specifically as follows:

[0095] The reaction force-displacement curve composed of the reaction force-displacement discrete data is used to obtain the closing moment of the contact in combination with the step point of the reaction force-displacement discrete data. The closing moment of the contact is corresponded to the reaction force-displacement curve to obtain the current contact opening distance. Finally, the current contact opening distance is subtracted from the initial contact opening distance to obtain the current contact wear value.

[0096] In this embodiment, combined with the above-mentioned process of obtaining the reaction force and the step point, the calculation principle diagram of the contact wear value is as follows: Figure 4 shown.

[0097] As a preferred implementation of this embodiment, the step of predicting the remaining service life of the corresponding relay contact based on the historical wear value of each relay is specifically as follows:

[0098] A long short-term memory neural network is used as the prediction model. The wear value and wear time of each contact are combined to train the prediction model. The trained prediction model is used to predict the remaining service life of the contact.

[0099] In this embodiment, a training model of a long short-term memory neural network is constructed for each relay, and training is performed according to the historical operation data of each contactor, and the training data is continuously updated during the operation of the contactor. At the same time, the contact erosion process of the relay is relatively slow, and the single contact wear amount is very small. Therefore, it is not necessary to train the neural network every time the contact wear is observed. The data can be updated once after multiple observations to train the neural network.

[0100] Before each model training, the data needs to be updated first. The historical contact wear data of a certain contactor is extracted from the historical data of the cloud platform. These data are used to make predictions for the neural network model. The long short-term memory neural network (LSTM) is used to predict the remaining service life of the relay contacts. After the model prediction is completed, the contact wear of the relay in the future is predicted. If it is predicted that its life is about to end, an early warning is issued to guide workers to replace or maintain the relay contacts; if the contact life is still at a healthy level in the short term, the model stops predicting, waits for a certain period of time, collects data again, and makes predictions again.

[0101] From the characteristics of the prediction model, it can be seen that the accuracy of short-term prediction is higher than that of long-term prediction, so the accuracy of this prediction method is expected to be higher than that of existing models; and because this model uses the relay's own historical data to predict its own future trend, the problem of different relay operating conditions is included in the model training process, which can further improve the model prediction accuracy; finally, because the model is deployed in the cloud and the prediction frequency for a single relay is very low, the deployment cost of the algorithm can be reduced a lot.

[0102] Embodiment 2:

[0103] A relay predictive maintenance method based on contact wear, the specific steps comprising:

[0104] Collect circuit parameters when each relay is closed;

[0105] Based on the D'Alembert mechanical motion equation, the reaction force formula related to contact wear is rewritten, and the reaction force on the moving iron core during the closing process in each relay is calculated;

[0106] Identify the step point of the reaction force, i.e. the collision moment of the contacts in the relay, based on the least squares method;

[0107] The wear value of the corresponding contact is calculated using the reaction force on the moving iron core of each relay and the step point of the reaction force;

[0108] Build a wide-area Internet of Things based on Lora communication and 4G communication, and transmit the contact wear values ​​collected by all contactors to the cloud platform;

[0109] The remaining service life of the corresponding relay contacts is predicted based on the historical wear value of each relay.

[0110] As a preferred implementation of this embodiment, the reaction force formula related to contact wear rewritten based on the D'Alembert mechanical motion equation is specifically:

[0111]

[0112] In the formula, x is the displacement distance of the moving iron core, t is the displacement time, F X is the electromagnetic attraction, F f For the reaction force.

[0113] As a preferred implementation of this embodiment, the step of identifying the step point of the reaction force based on the least square method, that is, the collision moment of the contacts in the relay, is specifically as follows:

[0114] Assume a point as the step point of the original reaction force-displacement discrete data, generate a basis function containing the same step point, use the basis function to achieve the least squares optimal fitting of the original reaction force-displacement discrete data, and record the fitting error; find the point in the original reaction force-displacement discrete data where the fitting error of the original reaction force-displacement discrete data using the basis function is the smallest, and this point is the step point of the original reaction force-displacement discrete data;

[0115] The formula of the basis function is as follows:

[0116]

[0117] The constraints are:

[0118] x0≤w slip ≤x n ,z slip ≠z zlip1

[0119] In the formula, w is the function variable, w slip is the function segmentation point, z slip is the function step left value, z slip1 is the right value of the function step; p1~p6 are coefficients, x0, x n is the distribution range of the original reaction force-displacement discrete data.

[0120] As a preferred implementation of this embodiment, the step of calculating the wear value of the contact using the reaction force of the moving iron core and the reaction force step point is specifically as follows:

[0121] The reaction force-displacement curve composed of the reaction force-displacement discrete data is used to obtain the closing moment of the contact in combination with the step point of the reaction force-displacement discrete data. The closing moment of the contact is corresponded to the reaction force-displacement curve to obtain the current contact opening distance. Finally, the current contact opening distance is subtracted from the initial contact opening distance to obtain the current contact wear value.

[0122] As a preferred implementation of this embodiment, the step of predicting the remaining service life of the contact using the wear value of the contact is specifically as follows:

[0123] A long short-term memory neural network is used as the prediction model. The wear value and wear time of each contact are combined to train the prediction model. The trained prediction model is used to predict the remaining service life of the contact.

[0124] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A relay predictive maintenance system based on contact wear, characterized in that: It includes a contact wear parameter acquisition module, a reaction force calculation module, a reaction force step identification module, a contact wear value calculation module, a data transmission module and a contact life prediction module, wherein each relay is connected to a separate contact life prediction module, the reaction force calculation module, the reaction force step identification module and the contact wear value calculation module are deployed at the edge end, and the contact life prediction module is arranged on the cloud platform; Contact wear parameter collection module, collects circuit parameters when each relay is closed; The reaction force calculation module rewrites the reaction force formula related to contact wear based on the D'Alembert mechanical motion equation and calculates the reaction force on the moving iron core during the closing process of each relay; The reaction force step identification module identifies the step point of the reaction force, i.e. the collision moment of the contacts in the relay, based on the least square method; The contact wear value calculation module calculates the wear value of the corresponding contact by using the reaction force on the moving iron core of each relay and the step point of the reaction force; The data transmission module builds a wide-area Internet of Things based on Lora communication and 4G communication, and transmits the contact wear values ​​collected by all contactors to the cloud platform; The contact life prediction module predicts the remaining service life of the corresponding relay contact based on the historical wear value of each relay.

2. A relay predictive maintenance system based on contact wear according to claim 1, characterized in that: The reaction force formula related to contact wear rewritten based on the D'Alembert mechanical motion equation is specifically: In the formula, x is the displacement distance of the moving iron core, t is the displacement time, F X is the electromagnetic attraction, F f For the reaction force.

3. A relay predictive maintenance system based on contact wear according to claim 1, characterized in that: The step of identifying the step point of the reaction force based on the least square method, that is, the collision moment of the contacts in the relay, is specifically as follows: Assume a point as the step point of the original reaction force-displacement discrete data, generate a basis function containing the same step point, use the basis function to achieve the least squares optimal fitting of the original reaction force-displacement discrete data, and record the fitting error; use the three-point search method to find the point in the original reaction force-displacement discrete data where the fitting error of the original reaction force-displacement discrete data using the basis function is the smallest, and this point is the step point of the original reaction force-displacement discrete data; The formula of the basis function is as follows: The constraints are: x0≤w slip ≤x n ,With slip ≠of zlip1 In the formula, w is the function variable, w slip is the function segmentation point, z slip is the function step left value, z slip1 is the right value of the function step; p1~p6 are coefficients, x0, x n is the distribution range of the original reaction force-displacement discrete data.

4. A relay predictive maintenance system based on contact wear according to claim 1, characterized in that: The steps of calculating the contact wear value using the reaction force of the moving iron core and the reaction force step point are specifically as follows: The reaction force-displacement curve composed of the reaction force-displacement discrete data is used to obtain the closing moment of the contact in combination with the step point of the reaction force-displacement discrete data. The closing moment of the contact is corresponded to the reaction force-displacement curve to obtain the current contact opening distance. Finally, the current contact opening distance is subtracted from the initial contact opening distance to obtain the current contact wear value.

5. The relay predictive maintenance system based on contact wear according to claim 1, characterized in that: The steps of predicting the remaining service life of the corresponding relay contact based on the historical wear value of each relay are specifically as follows: A long short-term memory neural network is used as the prediction model. The wear value and wear time of each contact are combined to train the prediction model. The trained prediction model is used to predict the remaining service life of the contact.

6. A relay predictive maintenance method based on contact wear, characterized in that: The specific steps include: Collect circuit parameters when each relay is closed; Based on the D'Alembert mechanical motion equation, the reaction force formula related to contact wear is rewritten, and the reaction force on the moving iron core during the closing process in each relay is calculated; Identify the step point of the reaction force, i.e. the collision moment of the contacts in the relay, based on the least squares method; The wear value of the corresponding contact is calculated using the reaction force on the moving iron core of each relay and the step point of the reaction force; Build a wide-area Internet of Things based on Lora communication and 4G communication, and transmit the contact wear values ​​collected by all contactors to the cloud platform; The remaining service life of the corresponding relay contacts is predicted based on the historical wear value of each relay.

7. A relay predictive maintenance method based on contact wear according to claim 6, characterized in that: The reaction force formula related to contact wear rewritten based on the D'Alembert mechanical motion equation is specifically: In the formula, x is the displacement distance of the moving iron core, t is the displacement time, F X is the electromagnetic attraction, F f For the reaction force.

8. The method for predictive maintenance of relays based on contact wear according to claim 6, characterized in that: The step of identifying the step point of the reaction force based on the least square method, that is, the collision moment of the contacts in the relay, is specifically as follows: Assume a point as the step point of the original reaction force-displacement discrete data, generate a basis function containing the same step point, use the basis function to achieve the least squares optimal fitting of the original reaction force-displacement discrete data, and record the fitting error; use the three-point search method to find the point in the original reaction force-displacement discrete data where the fitting error of the original reaction force-displacement discrete data using the basis function is the smallest, and this point is the step point of the original reaction force-displacement discrete data; The formula of the basis function is as follows: The constraints are: x0≤w slip ≤x n ,With slip ≠of zlip1 In the formula, w is the function variable, w slip is the function segmentation point, z slip is the function step left value, z slip1 is the right value of the function step; p1~p6 are coefficients, x0, x n is the distribution range of the original reaction force-displacement discrete data.

9. A relay predictive maintenance method based on contact wear according to claim 8, characterized in that: The steps of calculating the contact wear value using the reaction force of the moving iron core and the reaction force step point are specifically as follows: The reaction force-displacement curve composed of the reaction force-displacement discrete data is used to obtain the closing moment of the contact in combination with the step point of the reaction force-displacement discrete data. The closing moment of the contact is corresponded to the reaction force-displacement curve to obtain the current contact opening distance. Finally, the current contact opening distance is subtracted from the initial contact opening distance to obtain the current contact wear value.

10. The relay predictive maintenance method based on contact wear according to claim 6, characterized in that: The steps of predicting the remaining service life of the corresponding relay contact based on the historical wear value of each relay are specifically as follows: A long short-term memory neural network is used as the prediction model. The wear value and wear time of each contact are combined to train the prediction model. The trained prediction model is used to predict the remaining service life of the contact.

Images