Breaker service life prediction system and method based on Internet of Things

Abstract

The invention discloses a circuit breaker service life prediction system and method based on the Internet of Things, and relates to the technical field of circuit breakers, and the system comprises an Internet of Things data collection module, a scene grouping module, a data quantification module, a key factor identification module, a service life prediction module and an early warning priority output module. The Internet of Things data acquisition module is used for acquiring environment data and working condition data; the scene grouping module is used for performing scene grouping on the circuit breakers from two dimensions of environmental conditions and working condition loads; the data quantification module analyzes the comprehensive life loss of each circuit breaker; the key factor identification module is used for outputting key factors of the corresponding scene groups; the life prediction module is used for constructing a scenarized life prediction model; key factor data collected by the Internet of Things in real time are substituted into the prediction model to calculate the residual life; and the early warning priority output module is used for carrying out early warning response when the model output result is smaller than the residual life early warning time limit, and carrying out priority ranking on all label factors contained in the same scene group.

Description

Technical Field

[0001] This invention relates to the field of technology, specifically to a circuit breaker lifespan prediction system and method based on the Internet of Things. Background Technology

[0002] As power systems develop towards intelligence and distributed systems, circuit breakers, as core protective devices ensuring power supply security, directly impact the efficiency of power grid operation and maintenance and the reliability of power supply through accurate lifespan prediction. In different application scenarios, circuit breakers face significantly different environmental conditions and operating loads, leading to distinct scenario-specific lifespan loss mechanisms. For example, circuit breakers in coastal areas are prone to accelerated insulation aging due to salt spray corrosion, circuit breakers in metallurgical plants experience accelerated contact erosion due to inrush currents, and data center circuit breakers suffer from mechanical structural fatigue due to high-frequency opening and closing.

[0003] As a core protection device in the power system, the service life of circuit breakers directly affects the stability and security of power supply. Under different application scenarios, the environmental conditions and operating loads faced by circuit breakers vary significantly, leading to different lifespan loss mechanisms. Existing IoT-based circuit breaker lifespan prediction methods mostly rely on general data models, which have two major problems: First, they fail to design analysis logic for scenario-specific characteristics, ignoring scenario-specific influencing factors, resulting in large prediction errors. Second, they lack in-depth processing of IoT-collected data, resulting in low efficiency in identifying key factors and an inability to quickly locate core loss factors in different scenarios, thus affecting the accuracy of lifespan prediction. Third, the early warning mechanism lacks priority management; when multiple circuit breakers in the same scenario group trigger early warnings simultaneously, it is impossible to accurately identify factors and equipment with a greater impact on lifespan, leading to unreasonable allocation of maintenance resources and affecting intervention efficiency.

[0004] These issues make it difficult for existing technologies to accurately predict the lifespan of circuit breakers under different scenarios, which can easily lead to untimely or excessive maintenance, increase power grid operation and maintenance costs, and even cause equipment failures that result in power outages. Summary of the Invention

[0005] The purpose of this invention is to provide a circuit breaker lifespan prediction system and method based on the Internet of Things (IoT) to solve the problems raised in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting the service life of circuit breakers based on the Internet of Things, the method comprising the following steps: Step S100: Use IoT sensors deployed on the circuit breakers to collect environmental and operating data, and group the circuit breakers into scenarios from both environmental conditions and operating load dimensions; after assigning a scenario label to each circuit breaker, group the circuit breakers with label similarity greater than the similarity threshold into the same scenario group. Step S200: The mechanical life of each circuit breaker in the same scenario group is characterized by the frequency of opening and closing jamming, and the electrical life loss is characterized by the frequency of abnormal outlet temperature. Analyze the comprehensive life loss of each circuit breaker. Step S300: Standardize the label factors corresponding to the environmental data and operating condition data, analyze the correlation coefficients between each label factor and the comprehensive life loss within the same scenario group, and output the key factors of the corresponding scenario group. Step S400: Construct a scenario-based lifetime prediction model using key factors as input variables and comprehensive lifetime loss as output variables; and substitute the key factor data collected in real time by the Internet of Things into the prediction model to calculate the remaining lifetime. Step S500: Preset the remaining life warning time limit. When the model output result is less than the remaining life warning time limit, issue a warning response. When the circuit breaker operating condition changes abruptly, re-verify the remaining life. Based on the mathematical relationship between the remaining life after the change and the remaining life before the change, prioritize all the label factors included in the same scenario group.

[0007] The purpose of prioritization is to effectively prioritize intervention for circuit breakers that have a greater impact on remaining lifespan when multiple circuit breakers in the same scenario group respond to warnings due to different scenario factors at the same time.

[0008] Furthermore, step S100 includes the following specific processes: Environmental data include salt spray concentration, daily average temperature, dust concentration, and relative humidity; Operating data includes average monthly number of operations, peak current per single opening and closing, harmonic content, and continuous operating time; The IoT sensor sets the acquisition frequency and collects corresponding data with environmental data and operating condition data as the acquisition targets within the acquisition period, and assigns a scene tag value to each circuit breaker. For circuit breakers that have completed scene tag assignment, perform unified quantization of each tag. Unified quantization refers to converting the assigned values ​​of each scene tag into a quantization score within the range of 0-10 according to the corresponding quantization rules; calculate the total tag score Q for each circuit breaker, Q=∑q, where q represents the quantization score of each scene tag recorded by the circuit breaker; use the formula: S=[1-|Q A -Q B | / 80]×100%, calculate the label similarity between any two circuit breakers, where Q A Q represents the total score of the labels for circuit breaker A. B This indicates the total score of the label for circuit breaker B.

[0009] By transforming vague "scene features" into precise "numerical scores," the subjectivity of experience-based grouping is avoided, and scene division of circuit breakers in the same group is carried out to ensure the rationality of scene grouping. This provides a high-quality data foundation for subsequent key factor identification and model building. Moreover, the quantification rules can be automatically calculated directly based on real-time data collected by IoT sensors without manual intervention, which meets the core patent requirement of "intelligent analysis."

[0010] Furthermore, the analysis of the overall lifespan loss of each circuit breaker includes the following: Obtain the mechanical life loss L recorded for each circuit breaker. 机械 L 机械 = Frequency of opening and closing interruptions / 4, and the electrical life loss L recorded by the circuit breaker. 电气 L 电气 =Frequency of temperature anomalies / 2; Using the formula: L 总 =0.4*L 机械 +0.6*L 电气 Calculate the overall life loss L of each circuit breaker. 总 .

[0011] Furthermore, step S300 includes the following specific steps: Step S310: Take the daily average values ​​for salt spray concentration, temperature, and dust concentration in the environmental factors, and take the monthly total number of operations, the single average value for peak current, and the total harmonic distortion rate for the operating condition factors. Standardize all factor data to the range of 0-1. Step S320: Extract sample data of each circuit breaker in the same scenario group within the historical collection period. Each group of sample data contains a comprehensive life loss and eight factor values. Calculate the correlation coefficient r between each labeled factor and the corresponding comprehensive life loss. Iterate through all labeled factors to generate corresponding correlation coefficients and select labeled factors whose |r| > correlation coefficient threshold r0 as key factors.

[0012] Furthermore, step S400 includes the following specific steps: Step S410: Constructing a scenario-based lifetime prediction model refers to constructing an LSTM scenario-based lifetime prediction model that incorporates an attention mechanism. The model formula is as follows: L 总 =σ(W1·Attention(X1,X2,…,Xn)+W2·LSTM(X1,X2,…,Xn)+b); Where σ is the Sigmoid activation function, which limits the output value to the interval between 0 and 1; Attention( ) is the attention mechanism layer, LSTM( ) represents the Long Short-Term Memory network layer, W1 and W2 are weight matrices, b is the bias term; X1, X2, ..., Xn represent the 1st, 2nd, ..., nth standardized key factors in the same scene group; Attention(X) = Σ1 n α i X i α i =exp(s(X i )) / Σ1 n exp(s(X j )); where s(X i ) is the scoring function; s(X) j ) indicates removing X i Scoring is done on all key factors other than the corresponding critical factors; Step S420: Collect real-time key factor data, standardize it, input it into the output of the attention mechanism layer, and obtain the supplementary loss captured by the long short-term memory network layer to calculate the predicted comprehensive lifetime loss L. 总 and remaining lifetime T 剩余 =(1-L 总 ) / a, where a represents the annual wear rate calculated historically.

[0013] Furthermore, step S500 includes the following: The criteria for judging sudden changes in the operating conditions of a circuit breaker are as follows: Collect the real-time tag factor value Xt from IoT sensors, and the average value X over the past 12 hours. 0,t-12 The mutation rate P is calculated using the formula: P = |X| / (X|). t -X 0,t-12 ) / X max |, where X max This indicates the maximum value corresponding to the label factor; Set a mutation rate threshold P0. When P > P0, output that the corresponding label factor has mutated. Return to the scenario-based lifetime prediction model, increase the weight of the corresponding mutated label factor, adjust the time step parameter of the LSTM layer, substitute the label factor values, re-predict the overall lifetime loss, and calculate the remaining lifetime T. 突 .

[0014] Furthermore, step S500 also includes the following: Iterate through all operational condition mutation events, using each label factor as the classification target, generate a set of mutation events corresponding to the same classification target, and extract the predicted remaining lifetime T before the mutation occurred from the set of mutation events. 剩余 The remaining lifetime T calculated after the mutation. 突 Calculate the difference index Z, Z=T 剩余 -T突 Generate a set of difference indices for the set of mutation events corresponding to each labeled factor; the number of events contained in the set of mutation events corresponding to different labeled factors is the same; Using the formula: V=D1 / D0, the degree of mutation impact V of each difference index set is calculated, where D1 represents the number of mutation events with a difference index greater than the difference index threshold, and D0 represents the total number of events in the mutation event set. The larger the value of the degree of mutation impact, the greater the degree of change in lifespan prediction caused by the mutation of the label factor. The labeled factors are sorted in descending order of the magnitude of their impact from mutations to generate a priority sequence. Labeled factors that appear earlier in the priority sequence indicate that they will be given higher priority when multiple labeled factors respond to warnings simultaneously.

[0015] The IoT-based circuit breaker lifespan prediction system includes an IoT data acquisition module, a scenario grouping module, a data quantification module, a key factor identification module, a lifespan prediction module, and an early warning priority output module. The IoT data acquisition module is used to collect environmental and operating condition data; The scenario grouping module is used to group circuit breakers into scenarios based on both environmental conditions and operating load. The data quantification module is used to analyze the comprehensive life loss of each circuit breaker in the same scenario group by characterizing the mechanical life of the circuit breaker by the frequency of opening and closing jamming and the electrical life loss by characterizing the frequency of abnormal outlet temperature. The key factor identification module is used to analyze the correlation coefficients between each labeled factor and the overall lifespan loss within the same scenario group, and output the key factors for the corresponding scenario group. The lifespan prediction module is used to build scenario-based lifespan prediction models and to input key factor data collected in real time by the Internet of Things into the prediction model to calculate the remaining lifespan. The warning priority output module is used to preset the remaining life warning time limit. When the model output result is less than the remaining life warning time limit, a warning response is issued. When the circuit breaker operating condition changes abruptly, the remaining life is re-verified. Based on the mathematical relationship between the remaining life after the change and the remaining life before the change, all tag factors included in the same scenario group are prioritized.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention uses dual-dimensional labels of environment and working conditions to quantify scoring and similarity calculation, transforming ambiguous scene features into precise numerical indicators, thereby achieving objective grouping of circuit breaker scenarios and avoiding subjective errors caused by human experience. 2. This invention analyzes the correlation coefficients between standardized labeling factors and overall lifetime loss to identify scenario-specific key factors with strong correlations, while eliminating weakly correlated irrelevant variables to avoid model redundancy. By combining an LSTM model with an attention mechanism, it dynamically allocates the weights of key factors through the attention mechanism, highlighting the influence of core factors, and captures the nonlinear loss patterns of key factors over time through the LSTM network, significantly improving the accuracy and adaptability of lifetime prediction under different scenarios.

[0017] 3. This invention prioritizes labeled factors within a scenario group by calculating the difference index of remaining lifespan before and after a sudden change in operating conditions and the degree of impact of the change, thus clarifying the weight of different factors on lifespan. When multiple circuit breakers trigger warnings simultaneously, factors and equipment with a greater impact can be addressed first, achieving a reasonable allocation of maintenance resources, improving intervention efficiency, reducing power grid maintenance costs, and ensuring the safe and stable operation of the power system. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the circuit breaker lifespan prediction method based on the Internet of Things of the present invention. Detailed Implementation

[0019] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0020] Example: Figure 1 As shown, this invention provides a method for predicting the service life of circuit breakers based on the Internet of Things (IoT), the method comprising the following steps: Step S100: Use IoT sensors deployed on the circuit breakers to collect environmental and operating data, and group the circuit breakers into scenarios from both environmental conditions and operating load dimensions; after assigning a scenario label to each circuit breaker, group the circuit breakers with label similarity greater than the similarity threshold into the same scenario group. Step S200: The mechanical life of each circuit breaker in the same scenario group is characterized by the frequency of opening and closing jamming, and the electrical life loss is characterized by the frequency of abnormal outlet temperature. Analyze the comprehensive life loss of each circuit breaker. Step S300: Standardize the label factors corresponding to the environmental data and operating condition data, analyze the correlation coefficients between each label factor and the comprehensive life loss within the same scenario group, and output the key factors of the corresponding scenario group. Step S400: Construct a scenario-based lifetime prediction model using key factors as input variables and comprehensive lifetime loss as output variables; and substitute the key factor data collected in real time by the Internet of Things into the prediction model to calculate the remaining lifetime. Step S500: Preset the remaining life warning time limit. When the model output result is less than the remaining life warning time limit, issue a warning response. When the circuit breaker operating condition changes abruptly, re-verify the remaining life. Based on the mathematical relationship between the remaining life after the change and the remaining life before the change, prioritize all the label factors included in the same scenario group.

[0021] The purpose of prioritization is to effectively prioritize intervention for circuit breakers that have a greater impact on remaining lifespan when multiple circuit breakers in the same scenario group respond to warnings due to different scenario factors at the same time.

[0022] Step S100 includes the following specific processes: Environmental data include salt spray concentration, daily average temperature, dust concentration, and relative humidity; Operating data includes average monthly number of operations, peak current per single opening and closing, harmonic content, and continuous operating time; The IoT sensor sets the acquisition frequency and collects corresponding data with environmental data and operating condition data as the acquisition targets within the acquisition period, and assigns a scene tag value to each circuit breaker. For circuit breakers that have completed scene tag assignment, perform unified quantization of each tag. Unified quantization refers to converting the assigned values ​​of each scene tag into a quantization score within the range of 0-10 according to the corresponding quantization rules; calculate the total tag score Q for each circuit breaker, Q=∑q, where q represents the quantization score of each scene tag recorded by the circuit breaker; use the formula: S=[1-|Q A -Q B | / 80]×100%, calculate the label similarity between any two circuit breakers, where Q A Q represents the total score of the labels for circuit breaker A. B This represents the total tag score for circuit breaker B. For the same scenario group, a tag index based on key factors will be generated. For example, if the key factors in scenario group 1 are salt spray concentration, temperature, average monthly operation, and peak current, then all circuit breakers in the same scenario group will be searched for the corresponding value of that factor to form the tag index. For instance, the tag for a circuit breaker in a coastal wind power substation would be "Salt spray concentration ≥45μg / m³ + Temperature 18-28℃ + Average monthly operation ≤60 times + Peak current ≤12kA". The purpose of building the tag index is to quickly classify different scenario groups. As shown in the example: the scene label includes salt spray concentration, assigned a value of 45ug / m³. 3 The temperature is 20 degrees Celsius, and the average number of operations is 30 per month. Let's take three labels as an example, but it actually includes eight categories of labels. The quantification score rules corresponding to the above three types of labels can be defined as follows: Salt spray concentration: 0-10μg / m³=10 points, 11-20=8 points, 21-30=6 points, 31-40=4 points, 41-50=2 points, ≥51=0 points; the lower the score, the more severe the salt spray. Temperature: 20-28℃ = 10 points, 10-19 = 8 points, 29-35 = 8 points, 5-9 = 6 points, 36-40 = 6 points, <5 or >40 = 0 points; Average monthly performance: ≤50 = 10 points, 51-100 = 8 points, 101-200 = 6 points, 201-500 = 4 points, ≥501 = 2 points; The quantified score of each circuit breaker in different scenario labels can be obtained, and the similarity threshold is generally set to 85%.

[0023] By transforming vague "scene features" into precise "numerical scores," the subjectivity of experience-based grouping is avoided, and scene division of circuit breakers in the same group is carried out to ensure the rationality of scene grouping. This provides a high-quality data foundation for subsequent key factor identification and model building. Moreover, the quantification rules can be automatically calculated directly based on real-time data collected by IoT sensors without manual intervention, which meets the core patent requirement of "intelligent analysis."

[0024] The analysis of the overall lifespan loss of each circuit breaker includes the following: Obtain the mechanical life loss L recorded for each circuit breaker. 机械 L 机械 = Frequency of opening and closing interruptions / 4, and the electrical life loss L recorded by the circuit breaker. 电气 L 电气 =Temperature anomaly frequency / 2; The reason for taking the denominator as 4 is that the denominator 4 corresponds to the threshold of the jamming frequency when "mechanical life loss reaches 70% (critical value)" - that is, the jamming frequency of opening and closing = 4 times / thousand operations; The reason for taking the denominator as 2 is that the denominator 2 corresponds to the threshold of the temperature anomaly frequency when "electrical life loss reaches 70% (critical value)" - that is, the anomaly frequency of contact temperature ≥75℃ = 2 times / day; Using the formula: L 总 =0.4*L 机械 +0.6*L 电气 Calculate the overall life loss L of each circuit breaker. 总 .

[0025] Step S300 includes the following specific steps: Step S310: Take the daily average values ​​for salt spray concentration, temperature, and dust concentration in the environmental factors, and take the monthly total number of operations, the single average value for peak current, and the total harmonic distortion rate for the operating condition factors. Standardize all factor data to the range of 0-1. Step S320: Extract sample data of each circuit breaker in the same scenario group within the historical collection period. Each group of sample data contains a comprehensive life loss and eight factor values. Calculate the correlation coefficient r between each labeled factor and the corresponding comprehensive life loss. Iterate through all labeled factors to generate corresponding correlation coefficients and select labeled factors whose |r| > correlation coefficient threshold r0 as key factors.

[0026] Step S400 includes the following specific steps: Step S410: Constructing a scenario-based lifetime prediction model refers to constructing an LSTM scenario-based lifetime prediction model that incorporates an attention mechanism. The model formula is as follows: L 总 =σ(W1·Attention(X1,X2,…,Xn)+W2·LSTM(X1,X2,…,Xn)+b); Where σ is the Sigmoid activation function, which limits the output value to the interval between 0 and 1; Attention( ) is the attention mechanism layer, LSTM( ) represents the Long Short-Term Memory network layer, W1 and W2 are weight matrices, b is the bias term; X1, X2, ..., Xn represent the 1st, 2nd, ..., nth standardized key factors in the same scene group; Attention(X) = Σ1 n α i X i α i =exp(s(X i )) / Σ1 n exp(s(X j )); where s(X i ) is the scoring function; s(X) j ) indicates removing X i Scoring is done on all key factors other than the corresponding critical factors; Step S420: Collect real-time key factor data, standardize it, input it into the output of the attention mechanism layer, and obtain the supplementary loss captured by the long short-term memory network layer to calculate the predicted comprehensive lifetime loss L. 总 and remaining lifetime T 剩余 =(1-L 总 ) / a, where a represents the annual wear rate calculated historically.

[0027] Step S500 includes the following: The criteria for judging sudden changes in the operating conditions of a circuit breaker are as follows: Collect the real-time tag factor value Xt from IoT sensors, and the average value X over the past 12 hours. 0,t-12 The mutation rate P is calculated using the formula: P = |X| / (X|).t -X 0,t-12 ) / X max |, where X max This indicates the maximum value corresponding to the label factor; Set a mutation rate threshold P0. When P > P0, output that the corresponding label factor has mutated. Return to the scenario-based lifetime prediction model, increase the weight of the corresponding mutated label factor, adjust the time step parameter of the LSTM layer, substitute the label factor values, re-predict the overall lifetime loss, and calculate the remaining lifetime T. 突 .

[0028] Step S500 also includes the following: Iterate through all operational condition mutation events, using each label factor as the classification target, generate a set of mutation events corresponding to the same classification target, and extract the predicted remaining lifetime T before the mutation occurred from the set of mutation events. 剩余 The remaining lifetime T calculated after the mutation. 突 Calculate the difference index Z, Z=T 剩余 -T 突 Generate a set of difference indices for the set of mutation events corresponding to each labeled factor; the number of events contained in the set of mutation events corresponding to different labeled factors is the same; Using the formula: V=D1 / D0, the mutation impact level V of each difference index set is calculated, where D1 represents the number of mutation events with a difference index greater than the difference index threshold, and D0 represents the total number of events in the mutation event set; for example, if the difference index is set to 2 months; the larger the value of the mutation impact level, the greater the impact of the mutation of the labeled factor on the degree of change in life expectancy prediction. The labeled factors are sorted in descending order of the magnitude of their impact from mutations to generate a priority sequence. Labeled factors that appear earlier in the priority sequence indicate that they will be given higher priority when multiple labeled factors respond to warnings simultaneously.

[0029] The IoT-based circuit breaker lifespan prediction system includes an IoT data acquisition module, a scenario grouping module, a data quantification module, a key factor identification module, a lifespan prediction module, and an early warning priority output module. The IoT data acquisition module is used to collect environmental and operating condition data; The scenario grouping module is used to group circuit breakers into scenarios based on both environmental conditions and operating load. The data quantification module is used to analyze the comprehensive life loss of each circuit breaker in the same scenario group by characterizing the mechanical life of the circuit breaker by the frequency of opening and closing jamming and the electrical life loss by characterizing the frequency of abnormal outlet temperature. The key factor identification module is used to analyze the correlation coefficients between each labeled factor and the overall lifespan loss within the same scenario group, and output the key factors for the corresponding scenario group. The lifespan prediction module is used to build scenario-based lifespan prediction models and to input key factor data collected in real time by the Internet of Things into the prediction model to calculate the remaining lifespan. The warning priority output module is used to preset the remaining life warning time limit. When the model output result is less than the remaining life warning time limit, a warning response is issued. When the circuit breaker operating condition changes abruptly, the remaining life is re-verified. Based on the mathematical relationship between the remaining life after the change and the remaining life before the change, all tag factors included in the same scenario group are prioritized.

[0030] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting the service life of circuit breakers based on the Internet of Things, characterized in that: The method includes the following steps: Step S100: Use IoT sensors deployed on the circuit breakers to collect environmental and operating data, and group the circuit breakers into scenarios from both environmental conditions and operating load dimensions; after assigning a scenario label to each circuit breaker, group the circuit breakers with label similarity greater than the similarity threshold into the same scenario group. Step S200: The mechanical life of each circuit breaker in the same scenario group is characterized by the frequency of opening and closing jamming, and the electrical life loss is characterized by the frequency of abnormal outlet temperature. Analyze the comprehensive life loss of each circuit breaker. Step S300: Standardize the label factors corresponding to the environmental data and operating condition data, analyze the correlation coefficients between each label factor and the comprehensive life loss within the same scenario group, and output the key factors of the corresponding scenario group. Step S400: Construct a scenario-based lifetime prediction model using key factors as input variables and comprehensive lifetime loss as output variables; and substitute the key factor data collected in real time by the Internet of Things into the prediction model to calculate the remaining lifetime. Step S500: Preset the remaining life warning time limit. When the model output result is less than the remaining life warning time limit, issue a warning response. When the circuit breaker operating condition changes abruptly, re-verify the remaining life. Based on the mathematical relationship between the remaining life after the change and the remaining life before the change, prioritize all the label factors included in the same scenario group.

2. The method for predicting the service life of circuit breakers based on the Internet of Things according to claim 1, characterized in that: Step S100 includes the following specific processes: The environmental data includes salt spray concentration, average daily temperature, dust concentration, and relative humidity; The operating data includes the average number of monthly operations, peak current of a single opening and closing operation, harmonic content, and continuous operating time. The IoT sensor is set to a collection frequency and collects corresponding data with environmental data and operating condition data as the collection targets within the collection period, and assigns a scene tag value to each circuit breaker. For circuit breakers that have completed scene tag assignment, perform unified quantization on each tag. The unified quantization refers to converting the assigned values ​​of each scene tag into a quantization score between 0 and 10 according to the corresponding quantization rules. Calculate the total tag score Q for each circuit breaker, Q = ∑q, where q represents the quantization score of each scene tag recorded by the circuit breaker. Use the formula: S = [1 - |Q] A -Q B | / 80]×100%, calculate the label similarity between any two circuit breakers, where Q A Q represents the total score of the labels for circuit breaker A. B This indicates the total score of the label for circuit breaker B.

3. The method for predicting the service life of circuit breakers based on the Internet of Things according to claim 1, characterized in that: The analysis of the overall lifespan loss of each circuit breaker includes the following: Obtain the mechanical life loss L recorded for each circuit breaker. 机械 L 机械 = Frequency of opening and closing interruptions / 4, and the electrical life loss L recorded by the circuit breaker. 电气 L 电气 =Frequency of temperature anomalies / 2; Using the formula: L 总 =0.4*L 机械 +0.6*L 电气 Calculate the overall life loss L of each circuit breaker. 总 .

4. The method for predicting the service life of circuit breakers based on the Internet of Things according to claim 1, characterized in that: Step S300 includes the following specific steps: Step S310: Take the daily average values ​​for salt spray concentration, temperature, and dust concentration in the environmental factors, and take the monthly total number of operations, the single average value for peak current, and the total harmonic distortion rate for the operating condition factors. Standardize all factor data to the range of 0-1. Step S320: Extract sample data of each circuit breaker in the same scenario group within the historical collection period. Each group of sample data contains a comprehensive life loss and eight factor values. Calculate the correlation coefficient r between each labeled factor and the corresponding comprehensive life loss. Iterate through all labeled factors to generate corresponding correlation coefficients and select labeled factors whose |r| > correlation coefficient threshold r0 as key factors.

5. The method for predicting the service life of circuit breakers based on the Internet of Things according to claim 1, characterized in that: Step S400 includes the following specific steps: Step S410: The construction of the scenario-based lifetime prediction model refers to the construction of an LSTM scenario-based lifetime prediction model that incorporates an attention mechanism. The model formula is as follows: L 总 =σ(W1·Attention(X1,X2,…,Xn)+W2·LSTM(X1,X2,…,Xn)+b); Where σ is the Sigmoid activation function, which limits the output value to the interval between 0 and 1; Attention( ) is the attention mechanism layer, LSTM( ) represents the Long Short-Term Memory network layer, W1 and W2 are weight matrices, b is the bias term; X1, X2, ..., Xn represent the 1st, 2nd, ..., nth standardized key factors in the same scene group; Attention(X)=Σ1 n a i X i ,a i =exp(s(X i )) / S1 n exp(s(X j )); among them s(X i ) is the scoring function; s(X) j ) indicates removing X i Scoring is done on all key factors other than the corresponding critical factors; Step S420: Collect real-time key factor data, standardize it, input it into the output of the attention mechanism layer, and obtain the supplementary loss captured by the long short-term memory network layer to calculate the predicted comprehensive lifetime loss L. 总 and remaining lifetime T 剩余 =(1-L 总 ) / a, where a represents the annual wear rate calculated historically.

6. The method for predicting the service life of circuit breakers based on the Internet of Things according to claim 1, characterized in that: Step S500 includes the following: The criteria for determining a sudden change in the operating condition of the circuit breaker are as follows: Collect the real-time tag factor value Xt from IoT sensors, and the average value X over the past 12 hours. 0,t-12 The mutation rate P is calculated using the formula: P = |X| / (X|). t -X 0,t-12 ) / X max |, where X max This indicates the maximum value corresponding to the label factor; Set a mutation rate threshold P0. When P>P0, output the corresponding label factor that has mutated. Returning to the scenario-based lifetime prediction model, the weights of corresponding mutation label factors are increased, and the time step parameters of the LSTM layer are adjusted. The label factor values ​​are then substituted to re-predict the overall lifetime loss and calculate the remaining lifetime T. 突 .

7. The method for predicting the service life of circuit breakers based on the Internet of Things according to claim 1, characterized in that: Step S500 also includes the following: Iterate through all operational condition mutation events, using each label factor as the classification target, generate a set of mutation events corresponding to the same classification target, and extract the predicted remaining lifetime T before the mutation occurred from the set of mutation events. 剩余 The remaining lifetime T calculated after the mutation. 突 Calculate the difference index Z, Z=T 剩余 -T 突 Generate a set of difference indices for the set of mutation events corresponding to each labeled factor; the number of events contained in the set of mutation events corresponding to different labeled factors is the same; The mutation impact V of each difference index set is calculated using the formula: V=D1 / D0, where D1 represents the number of mutation events with a difference index greater than the difference index threshold, and D0 represents the total number of events in the mutation event set. The label factors are sorted in descending order of their impact from the mutation, generating a priority sequence.

8. A circuit breaker lifespan prediction system based on the Internet of Things, using the circuit breaker lifespan prediction method based on the Internet of Things as described in any one of claims 1-6, characterized in that, The system includes an IoT data acquisition module, a scene grouping module, a data quantification module, a key factor identification module, a lifespan prediction module, and an early warning priority output module. The IoT data acquisition module is used to collect environmental data and operating condition data; The scenario grouping module is used to group circuit breakers into scenarios from two dimensions: environmental conditions and operating load. The data quantification module is used to characterize the mechanical life of each circuit breaker in the same scenario group by the frequency of opening and closing jamming, and the electrical life loss by the frequency of abnormal outlet temperature, and to analyze the comprehensive life loss of each circuit breaker. The key factor identification module is used to analyze the correlation coefficient between each label factor and the overall life loss within the same scenario group, and output the key factors of the corresponding scenario group. The lifespan prediction module is used to construct a scenario-based lifespan prediction model and to input key factor data collected in real time by the Internet of Things into the prediction model to calculate the remaining lifespan. The warning priority output module is used to preset the remaining life warning time limit. When the model output result is less than the remaining life warning time limit, a warning response is issued. When the circuit breaker operating condition changes abruptly, the remaining life is re-verified. Based on the mathematical relationship between the remaining life after the change and the remaining life before the change, all tag factors included in the same scenario group are prioritized.

Images