Power equipment fault prediction method based on support vector regression

Abstract

The invention discloses a power equipment fault prediction method based on support vector regression, relates to the technical field of equipment faults, and can discover an electric arc concentration effect as soon as possible and send a danger instruction in time by monitoring the state and change of an electric arc in a circuit breaker in real time, thereby improving the operation safety of power equipment and further avoiding the spreading of faults. Accurate fault assessment: in combination with electrical and mechanical parameter analysis, a fault isolation coefficient Ggxs and a recovery influence coefficient Hyxs can be further scientifically calculated, so that the fault assessment is more accurate, and an important basis can be provided for maintenance decision making; and support vector regression model optimization prediction: a fault prediction model is constructed by adopting a support vector regression technology, so that a complex relationship can be effectively processed, prediction is performed by combining a fault isolation coefficient and a recovery influence coefficient, an accurate fault prediction score Gfz is provided, and the accuracy and timeliness of fault early warning are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment failure, and in particular to a method for predicting power equipment failure based on support vector regression. Background Art

[0002] With the increasing complexity of power systems and the increase in equipment operating loads, the frequency and severity of power equipment failures are also on the rise. How to accurately predict power equipment failures and improve their reliability and safety has become an important research direction in current power engineering. In this field, support vector regression (SVR), as a common machine learning method, has been widely used in fault prediction and health management. Support vector regression can accurately predict the state changes of power equipment by constructing a regression model in a high-dimensional space. Therefore, the power equipment fault prediction method based on support vector regression aims to predict the potential failure risk of the equipment by accurately modeling the working state of the power equipment and combining the change information of various parameters.

[0003] At present, in the fault prediction of power equipment, traditional methods usually rely on the experience of maintenance personnel and regular inspection and maintenance, which has certain limitations. Through the service life of the equipment, appearance inspection and preset fault mode, the fault of the equipment is often discovered only when it occurs suddenly, which increases the difficulty and cost of repair. In complex equipment such as multi-circuit circuit breakers, the working conditions of power equipment (such as circuit breakers) are complex, and the occurrence of faults is often sudden and hidden, which is difficult to predict in time through traditional methods. For example, during the opening and closing process, the extinction state and generation risk of the arc are not effectively monitored and analyzed. Due to the lack of accurate real-time fault assessment, many faults are often discovered when the equipment is damaged or seriously abnormal, resulting in the risk of production interruption or equipment damage. If such faults are not discovered in time, they may cause the equipment to fail completely, thereby affecting the stability of the entire power system. For example, when the arc concentration effect of the circuit breaker occurs, if it is not identified in advance and appropriate measures are not taken, the arc may exist for a long time, causing contact damage, serious burning of electrical equipment, and even fire and other serious safety accidents. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method for predicting power equipment faults based on support vector regression, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented by the following technical scheme: a method for predicting power equipment faults based on support vector regression, comprising the following steps: S1. Real-time monitoring of the working state of the opening and closing of the circuit breaker, so as to monitor the arc state in the arc extinguishing chamber of the circuit breaker during the opening and closing process of the circuit breaker, so as to obtain the relevant formation change data information in each area of ​​the arc, and based on the relevant formation change data information in each area of ​​the arc, determine whether there is a concentration effect of the arc, and if so, issue a danger instruction; S2. Receive danger instructions, monitor relevant electrical parameters and relevant mechanical parameters during the opening and closing process, and analyze the extinction of the arc during the opening process and the risk of arc generation during the closing process based on the relevant electrical parameters and relevant mechanical parameters, so as to calculate the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs; S3. Use support vector regression technology to build a fault assessment prediction model, combine the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs, input the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs into the fault assessment prediction model, and after dimensionless processing, fit and output the fault prediction score Gfz; S4. Pre-set qualified range , by comparing the fault prediction score Gfz with the qualified range Conduct comparative analysis to comprehensively judge the damage of the current circuit breaker and take corresponding early warning measures based on the judgment results.

[0006] Preferably, the specific steps of S1 include: S11. A limit switch is set on the movable part of the contact in the circuit breaker in advance to feedback the fault status of opening and closing, and the feedback content is transmitted to the control center through the circuit. After receiving the feedback content, the control center classifies the feedback content to classify the opening state and the closing state; S12. After performing the classification operation, the arc state in the arc extinguishing chamber in the circuit breaker is monitored in real time, and the arc is divided into regions to obtain the arc anode region, arc column region and arc cathode region respectively, and the arc regions are marked, and the relevant formation change data information in each arc region is obtained based on the arc anode region, arc column region and arc cathode region, wherein the relevant formation change data information includes the temperature value in each region , contact gap and current density .

[0007] Preferably, the specific step S1 also includes: S13. According to the relevant formation change data information in each area of ​​the arc, the temperature value Wdz, the current density Dmd and the contact gap Cjj are correlated and dimensionlessly processed to calculate the presentation coefficient Cxxs. The presentation coefficient Cxxs is obtained by the following formula: ; In the formula, n represents the number of arc regions, i represents the number of each arc region, It is expressed as the temperature value in the i-th region, Expressed as the average temperature of the arc, is expressed as the current density in the ith region, Expressed as the average current density of the arc, Represented as the contact gap in the ith region, Expressed as the average contact gap of the arc, , and They are all weight values, and the specific values ​​are set by the user according to the situation.

[0008] Preferably, the specific step S1 also includes: S14, by comparing and analyzing the presentation coefficient Cxxs with the preset safety threshold Q, it is determined whether there is a concentration effect on the generation of the arc. The specific comparison content is as follows: If the presentation coefficient Cxxs exceeds the preset safety threshold Q, it will be judged that there is a concentrated effect of the arc generation, and a danger instruction will be triggered to further analyze the damage effect of the current arc on the circuit breaker; If the presentation coefficient Cxxs does not exceed the preset safety threshold Q, it will be determined that there is no concentration effect in the generation of the arc and the danger instruction will not be triggered.

[0009] Preferably, the specific steps of S2 include: S21, after receiving the danger command triggered by S14, monitor the relevant electrical parameters and relevant mechanical parameters during the opening and closing process, where the relevant electrical parameters include the breaking current during the opening process , arc voltage , arc extinguishing time , opening speed and current frequency , and the closing voltage during the closing process , Closing speed And the closing current ;Related mechanical parameters include the opening time at each break during the opening process And the opening time of each break during the closing process ; S22. Based on the relevant mechanical parameters in the opening process and the relevant mechanical parameters in the closing process, the opening timestamps and closing timestamps of the contacts at the several groups of breaks in the circuit breaker are extracted, and according to the opening timestamps and closing timestamps of the contacts at the several groups of breaks in the circuit breaker, the isolation differences and connection differences of the circuit breaker in different power circuits are analyzed respectively to obtain the opening action difference factor and closing action difference factor , which is obtained by the following formula: ; In the formula, m represents the number of fractures, j represents the corresponding fracture number, It is expressed as the opening time at the jth break. Expressed as the average opening time, It is expressed as the closing time at the jth break. Expressed as average closing time.

[0010] Preferably, the specific step S2 also includes: S23: Based on the relevant electrical parameters and mechanical parameters during the opening process, combined with the opening action difference factor obtained in S22 , analyze the extinction of arcs in different power circuits of the circuit breaker during the opening process, and obtain the fault isolation coefficient Ggxs after dimensionless processing. The fault isolation coefficient Ggxs is obtained by the following formula: ; In the formula, It is expressed as the arc extinguishing time. Expressed as breaking current, Expressed as arc voltage, Expressed as the current frequency, Expressed as the opening speed, , 2. 3. 4. 5 and 6 respectively represent the arc extinguishing time , breaking current , arc voltage , current frequency , Disconnection Action Difference Factor And opening speed The specific value is set by the user according to the situation.

[0011] Preferably, the specific step S2 also includes: S24: Based on the relevant electrical parameters and mechanical parameters during the closing process, combined with the closing action difference factor obtained in S22 , analyze the risk of arcing in different power circuits of the circuit breaker during the closing process, and obtain the recovery influence coefficient Hyxs after dimensionless processing. The recovery influence coefficient Hyxs is obtained by the following formula: ; In the formula, Expressed as the closing current, Expressed as the closing voltage, Expressed as closing speed, , , and Respectively expressed as closing current , Closing voltage , Closing action difference factor And closing speed The specific value is set by the user according to the situation.

[0012] Preferably, the specific steps of S3 include: S31. Use support vector regression technology to build an initial model, and use the relevant formation change data information in each area of ​​the arc, as well as the relevant electrical parameters and relevant mechanical parameters in the opening and closing process to train and test the initial model. Use the trained initial model as a state recognition model, obtain the feature information in the state recognition model, and train and test the state recognition model with the acquired feature information. Combined with the dangerous instructions triggered in S14, use the trained state recognition model as a fault assessment prediction model. After dimensionless processing, fit the output fault prediction score Gfz.

[0013] Preferably, the specific step S3 also includes: S32, the fault prediction score Gfz is obtained by the following formula: ; In the formula, and are weight values, Represents the correction constant, ln Expressed as a logarithmic function, where 0 < <1,0< <1, the specific value is set by the user according to the situation.

[0014] Preferably, the specific steps of S4 include: S41, by comparing the fault prediction score Gfz with the qualified range Compare and analyze to comprehensively judge the damage of the current circuit breaker. The specific judgment contents are as follows: If the fault prediction score Gfz falls within the qualified range At this time, it will be comprehensively judged that the current circuit breaker is in a safe state, and the operating state of the circuit breaker will continue to be monitored, and the currently acquired fault prediction score Gfz will be saved as historical data; If the fault prediction score Gfz does not fall within the qualified range At this time, it will be comprehensively judged that the current circuit breaker is not in a safe state, and the current circuit breaker will be temporarily disabled and switched to the backup equipment. At the same time, the fault alarm system will be activated, and a warning notification will be sent to the maintenance personnel, and multiple groups of circuit breakers will be arranged to work in parallel.

[0015] The present invention provides a method for predicting power equipment faults based on support vector regression, which has the following beneficial effects: (1) Real-time monitoring and early warning: By real-time monitoring of the arc state and its changes in the circuit breaker, the arc concentration effect can be discovered early and dangerous instructions can be issued in time, thereby improving the operating safety of power equipment and further avoiding the spread of faults; Accurate fault assessment: Combining electrical and mechanical parameter analysis, the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs can be further scientifically calculated, making the fault assessment more accurate and providing an important basis for maintenance decisions. Support vector regression model optimization prediction: The support vector regression technology is used to construct a fault prediction model, which can effectively handle complex relationships, combine the fault isolation coefficient and the recovery influence coefficient for prediction, and provide an accurate fault prediction score Gfz, which improves the accuracy and timeliness of fault warning. Accurate warning and efficient maintenance: By comparing the fault prediction score with the preset qualified range, a multi-faceted evaluation of the equipment damage is achieved, ensuring the timeliness and pertinence of the warning and maintenance measures, thereby optimizing the operation and maintenance management of the power equipment and further extending the service life of the equipment. In short, this method not only improves the fault prediction accuracy of power equipment, but also effectively reduces the potential losses and safety hazards caused by power equipment failures.

[0016] (2) This method associates the multi-dimensional parameters in each area of ​​the arc and calculates the presentation coefficient Cxxs after dimensionless processing. This multi-dimensional data processing method can comprehensively consider the temperature, strength and contact state of the arc in different areas, fully reflect the changes of the arc in different areas, and avoid the errors that may be caused by single parameter analysis. The dimensionless process ensures the balance of the influence of each parameter on the final result, thereby improving the accuracy of the fault prediction and evaluation model. By introducing the presentation coefficient Cxxs for quantitative analysis of the arc state, it is possible to effectively determine whether the arc has a concentration effect. The concentration effect reflects the degree of heat and energy accumulation in different areas of the arc, which is crucial for assessing the danger of the arc. If the presentation coefficient Cxxs exceeds the preset safety threshold Q, the system will promptly determine that the arc has a concentration effect and trigger a danger command. This concentration effect judgment mechanism enables the system to quickly detect potential dangers at the early stage of arc generation, take preventive measures in time, and reduce the risk of equipment damage and failure.

[0017] (3) The present invention can fully understand the working status of the circuit breaker in a multi-circuit situation by extracting the opening timestamp and closing timestamp of the contacts at each break in the circuit breaker. The advantage of this method is that it can monitor the opening and closing processes of each power circuit separately, identify the behavioral differences of each break during the operation process, and provide a basis for further analyzing the multi-circuit performance and isolation of the circuit breaker. This detailed timestamp analysis mechanism enables the system to detect possible operational inconsistencies between power circuits. Because excessive time differences will cause some breaks to bear too high current or arc energy, potential faults can be further avoided. By quantifying the difference factors and comparing them with preset standards, the system can effectively improve the efficiency of fault warning and equipment maintenance, and ensure the safety and reliability of power equipment.

[0018] (4) During the closing process, the present invention combines electrical and mechanical parameters such as closing current, closing voltage, closing speed, and closing action difference factors to analyze the risk of arcing in different power circuits. By calculating the recovery influence coefficient Hyxs, the system can quantitatively evaluate the closing risk of each circuit, identify possible arcing risks in advance, and reduce the possibility of damage to power equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The figure is a flow chart of a method for predicting faults of electric power equipment based on support vector regression according to the present invention. DETAILED DESCRIPTION

[0020] 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.

[0021] Example 1: Please refer to Figure 1 The present invention provides a method for predicting power equipment faults based on support vector regression, comprising the following steps: S1. Real-time monitoring of the working state of the opening and closing of the circuit breaker, so as to monitor the arc state in the arc extinguishing chamber of the circuit breaker during the opening and closing process of the circuit breaker, so as to obtain the relevant formation change data information in each area of ​​the arc, and based on the relevant formation change data information in each area of ​​the arc, determine whether there is a concentration effect of the arc, and if so, issue a danger instruction; S2. Receive danger instructions, monitor relevant electrical parameters and relevant mechanical parameters during the opening and closing process, and analyze the extinction of the arc during the opening process and the risk of arc generation during the closing process based on the relevant electrical parameters and relevant mechanical parameters, so as to calculate the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs; S3. Use support vector regression technology to build a fault assessment prediction model, combine the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs, input the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs into the fault assessment prediction model, and after dimensionless processing, fit and output the fault prediction score Gfz; S4. Pre-set qualified range , by comparing the fault prediction score Gfz with the qualified range Conduct comparative analysis to comprehensively judge the damage of the current circuit breaker and take corresponding early warning measures based on the judgment results.

[0022] In this embodiment, the method monitors the arc state in the circuit breaker in real time, discovers the arc concentration effect in time, and issues a danger command before the electrical equipment fails. By combining the relevant data in the arc area to form a change trend, and monitoring the electrical and mechanical parameters during the opening and closing process, the method improves the accuracy of early fault identification and enhances the fault warning capability. Compared with the traditional method, the present invention can take preventive measures in advance, reduce equipment damage and downtime, and improve the safety and reliability of power equipment. The present invention uses support vector regression technology to construct a fault assessment prediction model. As a powerful machine learning algorithm, SVR can perform efficient fitting and prediction in complex nonlinear relationships. By inputting the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs, fitting and outputting the fault prediction score Gfz after dimensionless processing, the model can provide accurate fault prediction scores according to different historical data and working conditions, thereby providing data support for judging the damage degree of the circuit breaker and taking corresponding maintenance measures. By comparing and analyzing the pre-set qualified range and the fault prediction score Gfz, the present invention provides an effective fault assessment method that can accurately judge the damage of the current circuit breaker in real time. Compared with the traditional periodic inspection method, the evaluation mechanism of the present invention is more intelligent and automated, reduces the interference of human factors, and provides a dynamic and real-time fault evaluation system. It can take corresponding early warning measures according to the scoring results, issue warnings in time, and prevent the fault from further expanding. Since potential faults can be discovered early, the system can take maintenance or isolation measures in advance, avoiding serious damage to the equipment due to faults, and reducing equipment downtime and maintenance costs caused by sudden faults. In addition, through intelligent monitoring and prediction, the present invention can reduce the frequency of manual inspections and effectively improve the operating efficiency of the system and the long-term stability of the equipment.

[0023] Example 2: Please refer to Figure 1 , specifically: S1 specific steps include: S11. A limit switch is set on the movable part of the contact in the circuit breaker in advance to feedback the fault status of opening and closing, and the feedback content is transmitted to the control center through the circuit. After receiving the feedback content, the control center classifies the feedback content to classify the opening state and the closing state; S12. After performing the classification operation, the arc state in the arc extinguishing chamber in the circuit breaker is monitored in real time, and the arc is divided into regions to obtain the arc anode region, arc column region and arc cathode region respectively, and the arc regions are marked, and the relevant formation change data information in each arc region is obtained based on the arc anode region, arc column region and arc cathode region, wherein the relevant formation change data information includes the temperature value in each region , contact gap and current density .

[0024] In this embodiment, the method pre-sets a limit switch on the movable part of the contact in the circuit breaker to provide real-time feedback on the fault state of the opening and closing. This real-time feedback of the fault state can not only reflect the working state of the circuit breaker in a timely manner, but also transmit information to the control center through the circuit, and further perform classification operations. By classifying the opening and closing states, the operation of the circuit breaker under different working states can be clearly distinguished, and the operating state where the fault occurs can be accurately identified, which effectively reduces maintenance errors and equipment damage caused by misjudgment and improves the fault diagnosis accuracy of the circuit breaker. After performing the classification operation, the method monitors the arc state in the arc extinguishing chamber in the circuit breaker in real time, and divides the arc into anode area, arc column area and cathode area. Through this precise arc area division, the monitoring of the arc process can be refined to ensure that the arc state of each area is effectively tracked. The precise division of the arc anode area, arc column area and cathode area helps to grasp the propagation and change of the arc in real time, and provides a more detailed basis for subsequent fault assessment. This method is based on the relevant formation change data information in each area of ​​the arc, and collects multiple parameters including temperature values, contact gaps and current density. These multi-dimensional data can fully reflect the state of the arc and its changing trend, which helps to accurately analyze the stability and danger of the arc. For example, the temperature value can reflect the thermal state of the arc, and the contact gap and current density can help to determine the intensity of the arc and whether there is an overload risk. The collection and analysis of this multi-dimensional data enhances the reliability and accuracy of the fault prediction model. Through the marking processing and data collection of each area of ​​the arc, the method can detect abnormal changes in time at the early stage of arc formation and respond quickly. Unlike the traditional global monitoring method, the present invention adopts a local area monitoring method, which can detect any subtle changes in the early stage of arc generation, further improving the system's sensitivity to potential faults, and providing early fault warnings to avoid damage or shutdown of equipment due to arc faults. By inputting data such as the temperature values, contact gap and current density of each arc area into the subsequent prediction model, the system can automatically analyze the stability and risk of the arc and monitor the working status of the circuit breaker in real time without human intervention. This intelligent and automated working mode improves the reliability of the equipment as much as possible and effectively reduces the workload of manual inspection and the risk of misjudgment.

[0025] Example 3: Please refer to Figure 1 Specifically: S1 includes the following steps: S13. According to the relevant formation change data information in each area of ​​the arc, the temperature value Wdz, the current density Dmd and the contact gap Cjj are correlated and dimensionlessly processed to calculate the presentation coefficient Cxxs. The presentation coefficient Cxxs is obtained by the following formula: ; In the formula, n represents the number of arc regions, i represents the number of each arc region, It is expressed as the temperature value in the i-th region, Expressed as the average temperature of the arc, is expressed as the current density in the ith region, Expressed as the average current density of the arc, Represented as the contact gap in the ith region, Expressed as the average contact gap of the arc, , and are weight values, where 0 < <1,0< <1,0< <1, the specific value is set by the user according to the situation.

[0026] Temperature values ​​in the above areas Monitor and obtain through temperature sensor; Contact gap It refers to the physical distance between the circuit breaker contacts (including fixed contacts and moving contacts). During the opening and closing process of the circuit breaker, the contact gap will change. The size of the contact gap affects the quality of current cutting off or connecting, and is directly related to the arc generation and extinguishing effect. Usually, a displacement sensor (such as a laser displacement sensor or a capacitive displacement sensor) can be used to monitor the change of the contact gap.

[0027] Current density Refers to the amount of current passing through a unit cross-sectional area. It is usually used to evaluate the pressure of conductive materials when current flows. Too high current density may cause overheating, damage to circuit breaker contacts, or too long arc. Current density can usually be calculated by measuring the current passing through the circuit through a current sensor (such as a Hall effect sensor). The specific steps of S1 also include: S14, by comparing and analyzing the presentation coefficient Cxxs with the preset safety threshold Q, it is determined whether there is a concentration effect on the generation of the arc. The specific comparison content is as follows: If the presentation coefficient Cxxs exceeds the preset safety threshold Q, it will be judged that there is a concentrated effect of the arc generation, and a danger instruction will be triggered to further analyze the damage effect of the current arc on the circuit breaker; If the presentation coefficient Cxxs does not exceed the preset safety threshold Q, it will be determined that there is no concentration effect in the generation of the arc and the danger instruction will not be triggered.

[0028] In this embodiment, by performing dimensionless processing on the temperature value, the contact gap and the current density and calculating the presentation coefficient Cxxs, the present invention can comprehensively analyze the multi-dimensional data in each area of ​​the arc. The relevant data in each area can be fully considered. This method can not only reflect the changes of a single variable, but also capture the interaction effects of the variables in the arc area. By integrating and calculating these key parameters, the present invention can more accurately evaluate the state and development trend of the arc, and further improve the accuracy and credibility of arc monitoring. By comparing and analyzing the presentation coefficient Cxxs with the pre-set safety threshold Q, the present invention can determine whether there is a concentration effect of the arc. If the presentation coefficient Cxxs exceeds the safety threshold Q, a dangerous instruction will be triggered, so that the risk of the arc can be warned and further evaluated in time. This judgment method based on comparison and analysis makes the discovery of the arc concentration effect more efficient and accurate. By judging the concentration effect, the system can identify the damage that the arc may cause to the circuit breaker in advance, prevent the equipment from having serious failures due to the arc concentration effect, and improve the safety of the equipment. In the whole process, the system monitors the formation changes of the arc in real time and continuously updates the data to ensure real-time reflection of the arc state. Unlike traditional periodic inspection methods, the present invention can continuously monitor the state of the arc and make dynamic responses based on real-time data. This continuous, real-time monitoring method can adjust and optimize the operation strategy at any time to ensure that preventive or response measures can be taken quickly when the arc state becomes abnormal, thereby improving the system's response speed and processing capabilities to faults.

[0029] Example 4: Please refer to Figure 1 , specifically: S2 specific steps include: S21, after receiving the danger command triggered by S14, monitor the relevant electrical parameters and relevant mechanical parameters during the opening and closing process, where the relevant electrical parameters include the breaking current during the opening process , arc voltage , arc extinguishing time , opening speed and current frequency , and the closing voltage during the closing process , Closing speed And the closing current ;Related mechanical parameters include the opening time at each break during the opening process And the opening time of each break during the closing process ; S22. Based on the relevant mechanical parameters in the opening process and the relevant mechanical parameters in the closing process, the opening timestamps and closing timestamps of the contacts at several groups of breaks in the circuit breaker are extracted, and according to the opening timestamps and closing timestamps of the contacts at several groups of breaks in the circuit breaker, the isolation differences and connection differences of the circuit breaker in different power circuits (multiple breaks) are analyzed respectively to obtain the opening action difference factor and closing action difference factor , which is obtained by the following formula: ; In the formula, m represents the number of fractures, j represents the corresponding fracture number, It is expressed as the opening time at the jth break. Expressed as the average opening time, It is expressed as the closing time at the jth break. Expressed as average closing time.

[0030] The above mentioned tripping action difference factor and closing action difference factor It is used to quantify the discrete degree of closing and opening of each break in time, and further reflect the closing and opening performance of the circuit breaker; The opening time of each break during the above opening process And the opening time of each break during the closing process Both can be measured and obtained through time sensors (such as high-precision timers).

[0031] Different power circuits usually mean that each power circuit corresponds to an independent break. In a multi-break circuit breaker, each break carries a current circuit. When the circuits in the power system need to be opened or closed, the current of each circuit will be processed through its own break. In a multi-break circuit breaker, a break usually corresponds to a power circuit, and the function of each break is to independently open or close a current circuit.

[0032] In this embodiment, the present invention can quantify the action difference of each break in the opening and closing process by recording and analyzing the opening and closing time of each break in the circuit breaker in detail. By analyzing the isolation difference and connection difference between different breaks (i.e., different power circuits), the opening and closing performance of each circuit can be accurately evaluated. This analysis helps to find the action delay or inconsistency of different circuits, and then provide a basis for fault diagnosis and equipment optimization, ensuring the reliability of the system in a complex working environment. By extracting the contact opening timestamp and closing timestamp at each break, and using the formula to calculate the opening action difference factor and the closing action difference factor, the time discreteness of each break in the opening and closing process can be quantified. This method can effectively evaluate the opening and closing time difference of each break in the power circuit and identify potential performance imbalance problems.

[0033] Example 5: Please refer to Figure 1 Specifically: S2 includes the following specific steps: S23: Based on the relevant electrical parameters and mechanical parameters during the opening process, combined with the opening action difference factor obtained in S22 , analyze the extinction of arcs in different power circuits of the circuit breaker during the opening process, and obtain the fault isolation coefficient Ggxs after dimensionless processing. The fault isolation coefficient Ggxs is obtained by the following formula: ; In the formula, It is expressed as the arc extinguishing time. Expressed as breaking current, Expressed as arc voltage, Expressed as the current frequency, Expressed as the opening speed, , 2. 3. 4. 5 and 6 respectively represent the arc extinguishing time , breaking current , arc voltage , current frequency , Disconnection Action Difference Factor And opening speed The weight value of <1,0< 2<1,0< 3<1,0< 4<1,0< 5<1,0< 6<1, the specific value is set by the user according to the situation.

[0034] The arc extinguishing time mentioned above It refers to the time required from the time the current is cut off through the circuit breaker to the time the arc is completely extinguished. Usually, the presence and disappearance of the arc can be monitored by an arc sensor (such as a photoelectric sensor or an infrared sensor), and the arc extinguishing time can be obtained by monitoring the duration of the arc; Breaking current It refers to the current intensity cut off by the circuit breaker during the opening process. The breaking current determines the characteristics of the arc. Excessive breaking current may cause the arc to fail to extinguish. The current sensor can be used to measure the current and monitor the change of the current. Arc voltage It refers to the voltage value when an arc exists. The arc voltage is usually high. During the arc extinction process, the arc voltage will gradually decrease. Its value can be monitored and obtained by a voltage sensor. Current frequency It refers to the rate of change of current, indicating the frequency of current fluctuation. During the opening or closing process, the frequency change of current will affect the characteristics of the arc; its value can be measured by a frequency sensor or a current sensor to measure the frequency of the current and monitor the fluctuation of the current.

[0035] Opening speed It refers to the speed at which the contacts of the circuit breaker open during the opening process, which can be measured by a displacement sensor or a speed sensor.

[0036] The specific steps of S2 also include: S24: Based on the relevant electrical parameters and mechanical parameters during the closing process, combined with the closing action difference factor obtained in S22 , analyze the risk of arcing in different power circuits of the circuit breaker during the closing process, and obtain the recovery influence coefficient Hyxs after dimensionless processing. The recovery influence coefficient Hyxs is obtained by the following formula: ; In the formula, Expressed as the closing current, Expressed as the closing voltage, Expressed as closing speed, , , and Respectively expressed as closing current , Closing voltage , Closing action difference factor And closing speed The weight value of <1,0< <1,0< <1,0< <1, the specific value is set by the user according to the situation.

[0037] The above closing current It refers to the current that the circuit breaker allows to flow during the closing process. The size of the closing current will affect the generation and extinction of the arc during the closing process, and can be monitored and obtained by the current sensor; Closing voltage Refers to the voltage applied to the circuit when the circuit breaker is closed. The voltage will affect the impact of the closing current and the generation of arcs. It can be monitored and obtained by a voltage sensor. Closing speed It refers to the speed at which the contacts of the circuit breaker close during the closing process, which can be measured by a displacement sensor or a speed sensor.

[0038] In this embodiment, the present invention calculates the fault isolation coefficient Ggxs by analyzing the arc extinction of the circuit breaker in different power circuits during the opening process, combining relevant electrical and mechanical parameters and the opening action difference factor. This analysis method provides data support for evaluating the efficiency of arc extinction and the reliability of circuit isolation, especially in multi-break circuit breakers, and can accurately identify the arc extinction state to ensure the safety and stability of electrical equipment. Through dimensionless processing, the dimensional differences between different parameters can be eliminated, so that the data are comparable and consistent, thereby ensuring that the calculation of the fault isolation coefficient is more scientific and accurate. By calculating the fault isolation coefficient Ggxs, the present invention can effectively evaluate the isolation effect of different circuit breaks in the power line, provide an accurate basis for the stable operation and fault warning of electrical equipment, and effectively improve the safety and reliability of the power system. In short, by analyzing the arc extinction and the risk of arc generation, the system provides an efficient and real-time monitoring and evaluation tool for the power circuit, which can timely discover and respond to potential risks and improve the safety and stability of the power system.

[0039] Example 6: Please refer to Figure 1 , specifically: S3 specific steps include: S31. Use support vector regression technology to build an initial model, and use the relevant formation change data information in each area of ​​the arc, as well as the relevant electrical parameters and relevant mechanical parameters in the opening and closing process to train and test the initial model. Use the trained initial model as a state recognition model, obtain the feature information in the state recognition model, and train and test the state recognition model with the acquired feature information. Combined with the dangerous instructions triggered in S14, use the trained state recognition model as a fault assessment prediction model. After dimensionless processing, fit the output fault prediction score Gfz.

[0040] The specific steps of S3 also include: S32, the fault prediction score Gfz is obtained by the following formula: ; In the formula, and are weight values, Represents the correction constant, ln Expressed as a logarithmic function, where 0 < <1,0< <1, the specific value is set by the user according to the situation.

[0041] The specific steps of S4 include: S41, by comparing the fault prediction score Gfz with the qualified range Compare and analyze to comprehensively judge the damage of the current circuit breaker. The specific judgment contents are as follows: If the fault prediction score Gfz falls within the qualified range At this time, it will be comprehensively judged that the current circuit breaker is in a safe state, and the operating status of the circuit breaker will continue to be monitored to ensure that the equipment is under normal conditions. Regular inspection and maintenance, such as checking contact wear, lubrication system and insulation conditions, will be carried out, and the currently obtained fault prediction score Gfz will be saved as historical data for subsequent analysis. At the same time, relevant data will be recorded and archived for long-term trend analysis; If the fault prediction score Gfz does not fall within the qualified range At this time, it will be comprehensively judged that the current circuit breaker is not in a safe state, and the current circuit breaker will be temporarily disabled and switched to the backup equipment to prevent major failures. At the same time, the fault alarm system will be activated, and a warning notification will be sent to the maintenance personnel, and multiple groups of circuit breakers will be arranged to work in parallel to further ensure the continuous operation of the system and avoid the complete failure of the system due to the subsequent failure of a single circuit breaker.

[0042] The above qualified scope The setting method is as follows: Based on historical data, several groups of fault prediction scores Gfz are obtained, and combined with statistical algorithms, the average fault prediction scores are calculated respectively. and the standard deviation of the failure prediction score , and based on the mean failure prediction score and the standard deviation of the failure prediction score , set the qualified range: ;in, is a constant, usually with a value of 1-3, corresponding to different confidence levels, and the specific value is set by the user (according to actual conditions). Similarly, the setting method of the above safety threshold Q is as follows: The settings are similar.

[0043] In this embodiment, the present invention constructs an initial model by using support vector regression (SVR) technology, and combines the electrical and mechanical parameters in the process of opening and closing for training and testing, which can effectively obtain the change information in each area of ​​the arc. The establishment of this model makes the fault prediction of the circuit breaker more accurate. Through dimensionless processing, the trained state recognition model can fit and output the fault prediction score Gfz, which can be used to measure the current state of the circuit breaker. This method further improves the accuracy of fault prediction and helps to identify potential problems in time, thereby preventing the performance of the equipment from degrading before the fault occurs. Through the multi-dimensional data fusion method, not only the accuracy of the state recognition model is improved, but also the system can more comprehensively capture the precursors of fault risks, so as to more effectively predict and evaluate the operating state of the circuit breaker. When the fault prediction score falls into the qualified range, the system will judge that the circuit breaker is in a safe state, continue to monitor, and ensure that the equipment is maintained on schedule under normal conditions. This process not only ensures the continuous operation of the equipment, but also provides historical data support for long-term trend analysis. If the fault prediction score does not fall within the qualified range, the system can trigger a fault alarm in time, send an early warning notification to the maintenance personnel, and take emergency measures such as backup equipment switching and parallel operation to avoid the failure of the entire system due to a single circuit breaker failure. When the system detects that the fault prediction score of the circuit breaker deviates from the qualified range, the system will automatically start the early warning mechanism, disable the faulty circuit breaker in time and switch to the backup equipment. This mechanism effectively avoids the shutdown or catastrophic failure of the power system due to a single equipment failure. Through this intelligent decision-making process, the system can ensure the continuous operation of power facilities and the stability of power supply, and reduce the time of system shutdown and fault repair. In short, the fault prediction score Gfz of the system provides accurate equipment evaluation indicators, effectively prevents system shutdowns caused by equipment failures, and provides important support for equipment maintenance and management. Through adaptive parameter setting and historical data analysis, the system can continuously optimize operation and improve the safety, stability and reliability of the power system.

[0044] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for predicting power equipment faults based on support vector regression, characterized in that: The following steps are included: S1. Real-time monitoring of the working state of the opening and closing of the circuit breaker, so as to monitor the arc state in the arc extinguishing chamber of the circuit breaker during the opening and closing process of the circuit breaker, so as to obtain the relevant formation change data information in each area of ​​the arc, and based on the relevant formation change data information in each area of ​​the arc, determine whether there is a concentration effect of the arc, and if so, issue a danger instruction; S2. Receive danger instructions, monitor relevant electrical parameters and relevant mechanical parameters during the opening and closing process, and analyze the extinction of the arc during the opening process and the risk of arc generation during the closing process based on the relevant electrical parameters and relevant mechanical parameters, so as to calculate the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs; S3. Use support vector regression technology to build a fault assessment prediction model, combine the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs, input the fault isolation coefficient Ggxs and the recovery influence coefficient Hyxs into the fault assessment prediction model, and after dimensionless processing, fit and output the fault prediction score Gfz; S4. Pre-set qualified range , by comparing the fault prediction score Gfz with the qualified range Conduct comparative analysis to comprehensively judge the damage of the current circuit breaker and take corresponding early warning measures based on the judgment results.

2. The method for predicting power equipment faults based on support vector regression according to claim 1 is characterized in that: The specific steps of S1 include: S11. A limit switch is set on the movable part of the contact in the circuit breaker in advance to feedback the fault status of opening and closing, and the feedback content is transmitted to the control center through the circuit. After receiving the feedback content, the control center classifies the feedback content to classify the opening state and the closing state; S12. After performing the classification operation, the arc state in the arc extinguishing chamber in the circuit breaker is monitored in real time, and the arc is divided into regions to obtain the arc anode region, arc column region and arc cathode region respectively, and the arc regions are marked, and the relevant formation change data information in each arc region is obtained based on the arc anode region, arc column region and arc cathode region, wherein the relevant formation change data information includes the temperature value in each region , contact gap and current density .

3. The method for predicting power equipment faults based on support vector regression according to claim 2 is characterized in that: The specific steps of S1 also include: S13. According to the relevant formation change data information in each area of ​​the arc, the temperature value Wdz, the current density Dmd and the contact gap Cjj are correlated and dimensionlessly processed to calculate the presentation coefficient Cxxs. The presentation coefficient Cxxs is obtained by the following formula: ； In the formula, n represents the number of arc regions, i represents the number of each arc region, It is expressed as the temperature value in the i-th region, Expressed as the average temperature of the arc, is expressed as the current density in the ith region, Expressed as the average current density of the arc, Represented as the contact gap in the ith region, Expressed as the average contact gap of the arc, , and They are all weight values, and the specific values ​​are set by the user according to the situation.

4. The method for predicting power equipment faults based on support vector regression according to claim 3 is characterized in that: The specific steps of S1 also include: S14, by comparing and analyzing the presentation coefficient Cxxs with the preset safety threshold Q, it is determined whether there is a concentration effect on the generation of the arc. The specific comparison content is as follows: If the presentation coefficient Cxxs exceeds the preset safety threshold Q, it will be judged that there is a concentrated effect of the arc generation, and a danger instruction will be triggered to further analyze the damage effect of the current arc on the circuit breaker; If the presentation coefficient Cxxs does not exceed the preset safety threshold Q, it will be determined that there is no concentration effect in the generation of the arc and the danger instruction will not be triggered.

5. The method for predicting power equipment faults based on support vector regression according to claim 4 is characterized in that: The specific steps of S2 include: S21, after receiving the danger command triggered by S14, monitor the relevant electrical parameters and relevant mechanical parameters during the opening and closing process, where the relevant electrical parameters include the breaking current during the opening process , arc voltage , arc extinguishing time , opening speed and current frequency , and the closing voltage during the closing process , Closing speed And the closing current ;Related mechanical parameters include the opening time at each break during the opening process And the opening time of each break during the closing process ; S22. Based on the relevant mechanical parameters in the opening process and the relevant mechanical parameters in the closing process, the opening timestamps and closing timestamps of the contacts at the several groups of breaks in the circuit breaker are extracted, and according to the opening timestamps and closing timestamps of the contacts at the several groups of breaks in the circuit breaker, the isolation differences and connection differences of the circuit breaker in different power circuits are analyzed respectively to obtain the opening action difference factor and closing action difference factor , which is obtained by the following formula: ； In the formula, m represents the number of fractures, j represents the corresponding fracture number, It is expressed as the opening time at the jth break. Expressed as the average opening time, It is expressed as the closing time at the jth break. Expressed as average closing time.

6. The method for predicting power equipment faults based on support vector regression according to claim 5 is characterized in that: The specific steps of S2 also include: S23: Based on the relevant electrical parameters and mechanical parameters during the opening process, combined with the opening action difference factor obtained in S22 , analyze the extinction of arcs in different power circuits of the circuit breaker during the opening process, and obtain the fault isolation coefficient Ggxs after dimensionless processing. The fault isolation coefficient Ggxs is obtained by the following formula: ； In the formula, It is expressed as the arc extinguishing time. Expressed as breaking current, Expressed as arc voltage, Expressed as the current frequency, Expressed as the opening speed, , 2.

3.

4. 5 and 6 respectively represent the arc extinguishing time , breaking current , arc voltage , current frequency , Disconnection Action Difference Factor And opening speed The specific value is set by the user according to the situation.

7. The method for predicting power equipment faults based on support vector regression according to claim 5 is characterized in that: The specific steps of S2 also include: S24: Based on the relevant electrical parameters and mechanical parameters during the closing process, combined with the closing action difference factor obtained in S22 , analyze the risk of arcing in different power circuits of the circuit breaker during the closing process, and obtain the recovery influence coefficient Hyxs after dimensionless processing. The recovery influence coefficient Hyxs is obtained by the following formula: ； In the formula, Expressed as the closing current, Expressed as the closing voltage, Expressed as closing speed, , , and Respectively expressed as closing current , Closing voltage , Closing action difference factor And closing speed The specific value is set by the user according to the situation.

8. The method for predicting power equipment faults based on support vector regression according to claim 4 is characterized in that: The specific steps of S3 include: S31. Use support vector regression technology to build an initial model, and use the relevant formation change data information in each area of ​​the arc, as well as the relevant electrical parameters and relevant mechanical parameters in the opening and closing process to train and test the initial model. Use the trained initial model as a state recognition model, obtain the feature information in the state recognition model, and train and test the state recognition model with the acquired feature information. Combined with the dangerous instructions triggered in S14, use the trained state recognition model as a fault assessment prediction model. After dimensionless processing, fit the output fault prediction score Gfz.

9. The method for predicting power equipment faults based on support vector regression according to claim 8, characterized in that: The specific steps of S3 also include: S32, the fault prediction score Gfz is obtained by the following formula: ； In the formula, and are weight values, Represents the correction constant, ln Expressed as a logarithmic function, where 0 < <1,0< <1, the specific value is set by the user according to the situation.

10. The method for predicting power equipment faults based on support vector regression according to claim 1, characterized in that: The specific steps of S4 include: S41, by comparing the fault prediction score Gfz with the qualified range Compare and analyze to comprehensively judge the damage of the current circuit breaker. The specific judgment contents are as follows: If the fault prediction score Gfz falls within the qualified range At this time, it will be comprehensively judged that the current circuit breaker is in a safe state, and the operating state of the circuit breaker will continue to be monitored, and the currently acquired fault prediction score Gfz will be saved as historical data; If the fault prediction score Gfz does not fall within the qualified range At this time, it will be comprehensively judged that the current circuit breaker is not in a safe state, and the current circuit breaker will be temporarily disabled and switched to the backup equipment. At the same time, the fault alarm system will be activated, and a warning notification will be sent to the maintenance personnel, and multiple groups of circuit breakers will be arranged to work in parallel.

Images