A method for monitoring the state of power equipment based on biosensor technology

Abstract

The present invention relates to the technical field of power equipment status monitoring, and specifically, to a power equipment status monitoring method based on biosensor technology. It includes the following steps: S1, continuously monitoring and recording chemical change data of power equipment through biosensors, and continuously monitoring and recording electrical parameter change data during the operation of power equipment through electrical sensors; S2, analyzing electrical parameter change data through K-means algorithm to identify the health status pattern of power equipment; S3, based on the analysis results, evaluating the status of power equipment to determine whether there is a potential fault. The present invention is designed to use K-means algorithm combined with mutual information analysis between electrical features and target variables (such as equipment health status), which can more accurately identify the health status pattern of power equipment. This method can distinguish equipment with different health status.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment status monitoring, and in particular to a power equipment status monitoring method based on biosensor technology. Background Art

[0002] During long-term operation, various chemical changes may occur in power equipment, such as aging of oil and corrosion of metal parts. If these changes are not monitored, they may cause equipment performance degradation or even failure. Traditional monitoring methods may not be able to capture these subtle changes in time. During operation, power equipment will generate a large amount of electrical parameter data. Traditional analysis methods may not be accurate enough to identify the health status of the equipment. In summary, a power equipment status monitoring method based on biosensor technology is provided. Summary of the invention

[0003] The purpose of the present invention is to provide a method for monitoring the status of electric power equipment based on biosensor technology to solve the problems of low efficiency in real-time monitoring of the health status of equipment and low accuracy in pattern recognition of the health status of equipment proposed in the above-mentioned background technology.

[0004] To achieve the above object, the present invention provides a method for monitoring the state of electric power equipment based on biosensor technology, comprising the following steps:

[0005] S1. Continuously monitor and record chemical change data of power equipment through biosensors, and continuously monitor and record electrical parameter change data during the operation of power equipment through electrical sensors;

[0006] S2. Analyze the electrical parameter change data through the K-means algorithm to identify the pattern of the health status of the power equipment. In the process of calculating the mutual information between each electrical feature and the target variable, the vibration of the equipment, the equipment temperature, the metal particle concentration, the oxide concentration, the humidity and the weight of the target variable are introduced for optimization.

[0007] S3. Based on the analysis results, evaluate the status of the power equipment to determine whether there is a potential fault.

[0008] As a further improvement of the present technical solution, in S1, the chemical change data includes metal particle concentration and oxide concentration.

[0009] As a further improvement of the technical solution, in S2, the electrical parameter change data is analyzed by the K-means algorithm to identify the pattern of the health status of the power equipment, including the following steps:

[0010] S2.1. Preprocessing the electrical parameter change data;

[0011] S2.2, using wavelet transform to extract electrical features related to the health status of power equipment;

[0012] S2.3, select electrical features through mutual information method to identify the most critical electrical features for pattern recognition of health status of power equipment;

[0013] S2.4. Identify the pattern of health status of power equipment through K-means algorithm, and verify the identification results of K-means algorithm through cross-validation.

[0014] As a further improvement of the technical solution, in S2.2, the electrical characteristics related to the health status of the power equipment are extracted by using wavelet transform:

[0015] ;

[0016] A dynamic range compensation factor is introduced to optimize the process of wavelet transform to extract electrical features related to the health status of power equipment:

[0017] ;

[0018] in, Represents the result of continuous wavelet transform; Represents time series data; represents the scale parameter; represents the translation parameter; represents the time variable; represents the complex conjugate of the wavelet basis function; It represents the result of continuous wavelet transform by introducing dynamic range compensation factor; Indicates the dynamic range compensation factor.

[0019] As a further improvement of the technical solution, in S2.3, electrical feature selection is performed by mutual information method, including the following steps:

[0020] S2.31, preprocessing electrical characteristics;

[0021] S2.32, calculating the marginal probability distribution of each electrical feature and the marginal probability distribution of the target variable;

[0022] S2.33, calculating the mutual information between each electrical feature and the target variable, where the target variable indicates whether the electrical equipment is in a healthy state;

[0023] S2.34. Sort the electrical features according to the calculated mutual information values. The larger the mutual information value, the stronger the correlation between the electrical feature and the target variable.

[0024] S2.35. Set a threshold q. The electrical features whose mutual information is greater than the threshold are key features.

[0025] As a further improvement of the technical solution, in S2.33, the mutual information between each electrical feature and the target variable is calculated as follows:

[0026] ;

[0027] In the process of calculating the mutual information between each electrical feature and the target variable, the vibration of the device, the device temperature, the metal particle concentration, the oxide concentration, the humidity and the weight of the target variable are introduced for optimization:

[0028] ;

[0029] in, Indicates electrical characteristics and the target variable the degree of interdependence between them; represents a set of electrical characteristics; represents the target variable; Indicates electrical characteristics At a given value The probability distribution under Represents the target variable At a given value The probability distribution under Indicates electrical characteristics and the target variable At the same time, the value is and The joint probability distribution of Indicates a specific value of the monitored electrical characteristic; A specific value representing the device status; Representation variables At a given value The probability distribution under Indicates category The weight is used to deal with data imbalance; Representation variables At a given value Down, and At the same time, the value is and The joint probability of Representation variables At a given value Down, The value is The conditional probability of Representation variables At a given value Down, The value is The conditional probability of the variable Indicates equipment vibration, equipment temperature, metal particle concentration, oxide concentration and humidity; Representation variables The value of Indicates that the known variables In case of electrical characteristics and the target variable The degree of interdependence between them.

[0030] As a further improvement of the technical solution, in S2.4, the pattern of the health status of the power equipment is identified by the K-means algorithm, including the following steps:

[0031] S2.41, using the silhouette coefficient to determine the optimal number of clusters, and randomly selecting K electrical parameter change data as the initial centroids;

[0032] S2.42, clustering the electrical parameter change data by K-means algorithm;

[0033] S2.43. Identify different health status patterns of power equipment based on the clustering results, where each cluster represents a health status.

[0034] As a further improvement of the technical solution, in S2.41, the contour coefficient is specifically:

[0035] ;

[0036] in, Indicates the data point of electrical parameter change The average distance to other electrical parameter change data points in its cluster; Indicates the data point of electrical parameter change The average distance to all electrical parameter change data points in the nearest neighbor cluster; Represents the silhouette coefficient.

[0037] As a further improvement of the technical solution, in S2.42, the electrical parameter change data is clustered by a K-means algorithm, including the following steps:

[0038] S2.421. For each electrical parameter change data, find the nearest centroid and assign it to the corresponding cluster;

[0039] S2.422. For each cluster, calculate the new centroid;

[0040] S2.423. Repeat the steps until the center of mass no longer changes.

[0041] As a further improvement of the technical solution, in S3, based on the analysis result, the state of the power equipment is evaluated to determine whether there is a potential fault, including the following steps:

[0042] S3.1. According to the K-means algorithm analysis results in S2, a first cluster and a second cluster are obtained; the first cluster represents a normal operating mode of low resistance and stable current, and the second cluster represents a potential fault mode of high resistance and unstable current;

[0043] S3.2. Set a threshold m for the first cluster and a threshold n for the second cluster. The threshold is used to determine the normal operating state and the potential fault state. If the currently monitored data point is close to the centroid of the first cluster, the device is in a normal state; if the data point is close to the centroid of the second cluster, the device has a potential fault.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. In this power equipment status monitoring method based on biosensor technology, the K-means algorithm is used in combination with the mutual information analysis between electrical features and target variables (such as equipment health status) to more accurately identify the health status pattern of power equipment. This method can not only distinguish equipment with different health conditions, but also introduce weight optimization of parameters such as equipment vibration, temperature, metal particle concentration, oxide concentration, humidity, etc., which can further reduce the impact of interactions between features and data imbalance, thereby improving the accuracy of fault diagnosis.

[0046] 2. In the power equipment status monitoring method based on biosensor technology, by continuously monitoring the chemical change data of the power equipment (such as metal particle concentration, oxide concentration, etc.) and electrical parameter change data (such as resistance, current, power factor, etc.), it is possible to timely detect changes in the internal chemical composition of the equipment and abnormalities in the electrical characteristics. By analyzing the change trends of these data, possible problems with the equipment can be predicted in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The figure is a flow chart of the overall method of the present invention. DETAILED DESCRIPTION

[0048] 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. Example

[0049] See also Figure 1 As shown, this embodiment provides a method for monitoring the state of electric equipment based on biosensor technology, comprising the following steps:

[0050] S1. Continuously monitor and record chemical change data of power equipment through biosensors, and continuously monitor and record electrical parameter change data during the operation of power equipment through electrical sensors;

[0051] In this embodiment, the chemical change data includes metal particle concentration and oxide concentration. An increase in metal particle concentration will lead to an increase in contact surface resistance. An increase in oxide concentration will affect the stability of current. The presence of metal particles and oxides will lead to additional power loss.

[0052] S2, analyze the electrical parameter change data through K-means algorithm to identify the health status pattern of power equipment;

[0053] In this embodiment, the electrical parameter change data is analyzed by the K-means algorithm to identify the pattern of the health status of the power equipment, including the following steps:

[0054] S2.1. Preprocess the electrical parameter change data, including removing invalid data points and performing denoising on the data, such as filtering, to reduce the impact of measurement errors;

[0055] S2.2, using wavelet transform to extract electrical features related to the health status of power equipment, the electrical features including resistance, current and power factor;

[0056] Among them, wavelet transform provides a multi-resolution analysis tool that can observe signals at different scales. This means that it can capture both the global characteristics and local details of the signal, which is very useful for detecting transient phenomena that may exist in power equipment (such as short circuits, discharges, etc.). Since many faults in power equipment are manifested through transient signals, this ability is very important for early fault detection; wavelet transform can effectively separate useful information and noise in the signal. By selecting appropriate wavelet basis functions and threshold processing, the high-frequency noise part of the signal can be removed and the effective low-frequency information can be retained. This is very helpful for improving the accuracy of feature extraction; using wavelet transform to extract electrical features related to the health status of power equipment is specifically:

[0057] ;

[0058] In order to reduce the problem of electrical parameters changing too much or too little, the dynamic range compensation factor is introduced to optimize the process of wavelet transform to extract electrical features. The dynamic range compensation factor can adjust the dynamic range of the signal so that the signal with a large change amplitude can be compared with the signal with a small change amplitude on the same scale, thereby eliminating the feature extraction deviation caused by the difference in signal strength, which can ensure that the feature comparison between different signals is more fair and reasonable; through the dynamic range compensation factor, the response of the wavelet transform to signals of different strengths can be optimized, which helps to better capture the detailed features of the signal during the analysis process, especially when the signal strength changes drastically, it can still maintain a high feature extraction accuracy; when processing actual monitoring data, electrical parameters may fluctuate greatly due to different working conditions of the equipment. If these fluctuations are not processed, some weak but important signals may be masked in the feature extraction process. The dynamic range compensation factor can highlight those weak but key signal features, so that they can play their due role in pattern recognition; specifically:

[0059] ;

[0060] ;

[0061] in, Represents the result of continuous wavelet transform; Represents time series data, i.e., electrical parameter change data; Represents the scale parameter, that is, the degree of expansion and contraction of the wavelet function on the time axis; represents the translation parameter, that is, the position shift of the wavelet function on the time axis; represents the time variable; Represents the complex conjugate of the wavelet basis function. When the wavelet basis function is complex, the complex conjugate is used to ensure the real number nature of the transformation result. If it is a real wavelet basis function, no conjugate operation is required; It represents the result of continuous wavelet transform by introducing dynamic range compensation factor; Indicates the dynamic range compensation factor. Indicates the control of steepness; Indicates the average value of electrical parameter variation data;

[0062] S2.3, select electrical features through mutual information method to identify the most critical electrical features for pattern recognition of health status of power equipment;

[0063] Among them, mutual information is a statistic that measures the mutual dependence between two random variables. It can detect nonlinear relationships between variables. In the condition monitoring of power equipment, the relationship between electrical features and equipment health is often nonlinear. The mutual information method can effectively capture these nonlinear relationships, thereby selecting the most explanatory features for the health of the equipment; the mutual information method is not affected by the type of feature distribution. Whether it is a normal distribution or a skewed distribution, feature selection can be performed through mutual information. This is especially important for power equipment monitoring data, because these data usually contain multiple types of distributions and may contain outliers; by calculating the mutual information between electrical features and target variables (such as whether the equipment is healthy), features that are highly correlated with the target variable can be screened out, while excluding redundant features. This method helps to reduce the dimension of the feature space, improve the efficiency of model training and the accuracy of prediction; electrical feature selection through the mutual information method includes the following steps:

[0064] S2.31, preprocessing electrical characteristics, including data cleaning and data discretization;

[0065] S2.32, calculating the marginal probability distribution of each electrical feature and the marginal probability distribution of the target variable;

[0066] S2.33, calculating the mutual information between each electrical feature and the target variable, where the target variable indicates whether the electrical equipment is in a healthy state;

[0067] Furthermore, the mutual information between each electrical feature and the target variable is calculated as follows:

[0068] ;

[0069] In order to reduce the interaction between electrical features and data imbalance, the vibration of the equipment, equipment temperature, metal particle concentration, oxide concentration, humidity and the weight of the target variable are introduced for optimization in the process of calculating the mutual information between each electrical feature and the target variable. Equipment vibration can reflect the status of mechanical parts, such as bearing wear and imbalance, which are common causes of electrical failures. By monitoring vibration, these problems can be discovered in time to prevent them from further developing into serious electrical failures. Temperature is one of the important factors affecting the performance of electrical equipment. Too high or too low temperature will affect the electrical parameters of the equipment. By monitoring the temperature, the heat dissipation system can be adjusted in time to prevent equipment damage caused by excessive temperature. An increase in the concentration of metal particles is usually the result of mechanical wear, which may lead to electrical The resistance increases, affecting the reliability of electrical connections. By monitoring the concentration of metal particles, parts that need to be replaced or repaired can be identified in advance, so that preventive maintenance can be implemented to avoid electrical failures caused by wear. The increase in oxide concentration will affect the performance of insulating materials and may cause insulation failure, thereby affecting the stability of current. By monitoring the oxide concentration, the degree of aging of insulating materials can be evaluated. Humidity has a great impact on the insulation performance of electrical equipment. Excessive humidity may cause the insulation layer to absorb moisture, thereby reducing the insulation performance. Humidity monitoring can be used to warn of electrical failures caused by moisture, such as short circuits and leakage. By assigning different weights to target variables (such as health and fault status), the number of samples of different categories can be balanced when calculating mutual information to ensure that the model does not favor the features of the majority category. Specifically:

[0070] ;

[0071] ;

[0072] in, Indicates electrical characteristics and the target variable The degree of mutual dependence between the electrical characteristics and the target variable The mutual information between represents a set of electrical characteristics; represents the target variable; Indicates electrical characteristics At a given value The probability distribution under which the electrical characteristics The value is probability; Represents the target variable At a given value The probability distribution under Indicates electrical characteristics and the target variable At the same time, the value is and The joint probability distribution of The value is And the target variable The value is probability; Indicates a specific value of the monitored electrical characteristic; A specific value representing the device status; Representation variables At a given value The probability distribution under Indicates category The weight is used to deal with data imbalance; Representation variables At a given value Down, and At the same time, the value is and The joint probability of Representation variables At a given value Down, The value is The conditional probability of Representation variables At a given value Down, The value is The conditional probability of the variable Indicates equipment vibration, equipment temperature, metal particle concentration, oxide concentration and humidity; Representation variables The value of Indicates that the known variables In case of electrical characteristics and the target variable the degree of interdependence between them; Indicates that the device is vibrating; Indicates the device temperature; Indicates the concentration of metal particles; Indicates oxide concentration; Indicates humidity.

[0073] S2.34. Sort the electrical features according to the calculated mutual information values. The larger the mutual information value, the stronger the correlation between the electrical feature and the target variable.

[0074] S2.35, set a threshold q, the electrical features with mutual information greater than the threshold are key features;

[0075] S2.4, identify the pattern of health status of power equipment through K-means algorithm, and verify the identification results of K-means algorithm through cross-validation;

[0076] Among them, K-means is an unsupervised learning algorithm that can perform pattern recognition without pre-labeled data. This is very important for the status monitoring of power equipment, because the status data of the equipment during operation often does not have clear labels. The potential structure in the data can be automatically discovered in an unsupervised way; K-means can identify different patterns of the health status of power equipment and cluster similar data points together to form clusters. This pattern recognition capability helps to distinguish between normal operating conditions and potential fault conditions, thereby providing a basis for equipment maintenance; through the K-means algorithm, electrical features (such as resistance, current, power factor, etc.) can be clustered according to similarity, which helps to find which feature combinations can better characterize the health status of the equipment; Identifying patterns of the health status of power equipment through the K-means algorithm includes the following steps:

[0077] S2.41, using the silhouette coefficient to determine the optimal number of clusters, and randomly selecting K electrical parameter change data as the initial centroids;

[0078] Furthermore, the silhouette coefficient provides an objective indicator to evaluate the quality of clustering. It reflects the closeness of the data point to its cluster and the distance from other clusters. By maximizing the silhouette coefficient, the most reasonable number of clusters can be found to avoid the deviation of human subjective judgment. The silhouette coefficient is robust to outliers. By calculating the average distance between the data point and its cluster and the distance to the nearest neighbor cluster, the silhouette coefficient can better reflect the relationship between the data point and the cluster. Even if there are some outliers, it will not seriously affect the final clustering result. The value of the silhouette coefficient is between -1 and 1. The closer the value is to 1, the better the clustering effect is. The closer the value is to -1, the worse the clustering effect is. 0 means that the data point is on the boundary of the cluster. The intuitiveness of this numerical range helps to quickly understand the quality of the clustering results. The silhouette coefficient is specifically:

[0079] ;

[0080] in, Indicates the data point of electrical parameter change The average distance to other electrical parameter change data points in its cluster; Indicates the data point of electrical parameter change The average distance to all electrical parameter change data points in the nearest neighbor cluster (i.e., the closest cluster other than the cluster in which it is located); represents the silhouette coefficient;

[0081] S2.42, clustering the electrical parameter change data by K-means algorithm;

[0082] Furthermore, the electrical parameter change data is clustered by the K-means algorithm, including the following steps:

[0083] S2.421. For each electrical parameter change data, find the centroid closest to it and assign it to the corresponding cluster, where the centroid is the arithmetic mean of the coordinate values ​​of all data points in the cluster in each dimension;

[0084] S2.422. For each cluster, calculate the new centroid;

[0085] Among them, the new centroid ;

[0086] In the formula, Represents the index of the cluster; Representation Cluster The number of data points in the set size; Indicates the number of iterations; Represents the centroid, i.e., cluster The average position of all data points in ; Represents an index variable; Representation Cluster A data point in

[0087] S2.423. Repeat the steps until the center of mass no longer changes;

[0088] S2.43. Identify different health status patterns of power equipment based on the clustering results, where each cluster represents a health status.

[0089] S3. Based on the analysis results, evaluate the status of the power equipment to determine whether there is a potential fault;

[0090] In this embodiment, based on the analysis results, the state of the power equipment is evaluated to determine whether there is a potential fault, including the following steps:

[0091] S3.1. According to the K-means algorithm analysis results in S2, a first cluster and a second cluster are obtained; the first cluster represents a normal operating mode of low resistance and stable current, and the second cluster represents a potential fault mode of high resistance and unstable current;

[0092] S3.2, set a threshold m for the first cluster and a threshold n for the second cluster. The threshold is used to judge the normal operation state and the potential fault state. If the currently monitored data point is close to the centroid of the first cluster, the device is in a normal state; if the data point is close to the centroid of the second cluster, the device has a potential fault;

[0093] If two centroids have been obtained through the K-means algorithm and ; and set the threshold m to 0.5 and n to 0.8; the currently monitored data points ,in Represents resistance, Indicates current;

[0094] Compute the distance between two centroids:

[0095] ;

[0096] ;

[0097] in, represents the centroid of the normal operating mode of low resistance and steady current, Indicates the average resistance value in normal operating mode, Indicates the average current value in normal operating mode, The centroid of the potential failure mode representing high resistance, unstable current, represents the average resistance value under potential failure modes, Represents the average current value in potential failure mode, represents the distance between the centroids;

[0098] if , the device is in normal state, if , the device has a potential fault.

[0099] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A method for monitoring the state of power equipment based on biosensor technology, characterized in that: The following steps are involved: S1. Continuously monitor and record chemical change data of power equipment through biosensors, and continuously monitor and record electrical parameter change data during the operation of power equipment through electrical sensors; S2. Analyze the electrical parameter change data through the K-means algorithm to identify the pattern of the health status of the power equipment. In the process of calculating the mutual information between each electrical feature and the target variable, the vibration of the equipment, the equipment temperature, the metal particle concentration, the oxide concentration, the humidity and the weight of the target variable are introduced for optimization. S3. Based on the analysis results, evaluate the status of the power equipment to determine whether there is a potential fault; In S2, the electrical parameter change data is analyzed by the K-means algorithm to identify the pattern of the health status of the power equipment, including the following steps: S2.

1. Preprocessing the electrical parameter change data; S2.2, using wavelet transform to extract electrical features related to the health status of power equipment; S2.3, select electrical features through mutual information method to identify the most critical electrical features for pattern recognition of health status of power equipment; S2.4, identify the pattern of health status of power equipment through K-means algorithm, and verify the identification results of K-means algorithm through cross-validation; In S2.3, electrical feature selection is performed by mutual information method, including the following steps: S2.31, preprocessing electrical characteristics; S2.32, calculating the marginal probability distribution of each electrical feature and the marginal probability distribution of the target variable; S2.33, calculating the mutual information between each electrical feature and the target variable; S2.

34. Sort the electrical features according to the calculated mutual information values. The larger the mutual information value, the stronger the correlation between the electrical feature and the target variable. S2.35, set a threshold q, the electrical features with mutual information greater than the threshold are key features; In S2.33, the mutual information between each electrical feature and the target variable is calculated as follows: ； In the process of calculating the mutual information between each electrical feature and the target variable, the vibration of the device, the device temperature, the metal particle concentration, the oxide concentration, the humidity and the weight of the target variable are introduced for optimization: ； in, Indicates electrical characteristics and the target variable the degree of interdependence between them; represents a set of electrical characteristics; represents the target variable; Indicates electrical characteristics At a given value The probability distribution under Represents the target variable At a given value The probability distribution under Indicates electrical characteristics and the target variable At the same time, the value is and The joint probability distribution of Indicates a specific value of the monitored electrical characteristic; A specific value representing the device status; Representation variables At a given value The probability distribution under Indicates category The weight is used to deal with data imbalance; Representation variables At a given value Down, and At the same time, the value is and The joint probability of Representation variables At a given value Down, The value is The conditional probability of Representation variables At a given value Down, The value is The conditional probability of the variable Indicates equipment vibration, equipment temperature, metal particle concentration, oxide concentration and humidity; Representation variables The value of Indicates that the known variables In case of electrical characteristics and the target variable The degree of interdependence between them.

2. The method for monitoring the state of electric power equipment based on biosensor technology according to claim 1 is characterized in that: In S1, the chemical change data includes metal particle concentration and oxide concentration.

3. The method for monitoring the state of electric power equipment based on biosensor technology according to claim 2 is characterized in that: In S2.2, wavelet transform is used to extract electrical features related to the health status of power equipment, specifically: ； A dynamic range compensation factor is introduced to optimize the process of wavelet transform to extract electrical features related to the health status of power equipment: ； in, Represents the result of continuous wavelet transform; Represents time series data; represents the scale parameter; represents the translation parameter; represents the time variable; represents the complex conjugate of the wavelet basis function; It represents the result of continuous wavelet transform by introducing dynamic range compensation factor; Indicates the dynamic range compensation factor.

4. The method for monitoring the state of electric power equipment based on biosensor technology according to claim 3 is characterized in that: In S2.4, the pattern of the health status of the power equipment is identified by the K-means algorithm, including the following steps: S2.41, using the silhouette coefficient to determine the optimal number of clusters, and randomly selecting K electrical parameter change data as the initial centroids; S2.42, clustering the electrical parameter change data by K-means algorithm; S2.

43. Identify different health status patterns of power equipment based on the clustering results, where each cluster represents a health status.

5. The method for monitoring the state of electric power equipment based on biosensor technology according to claim 4 is characterized in that: In S2.41, the contour coefficient is specifically: ； in, Indicates the data point of electrical parameter change The average distance to other electrical parameter change data points in its cluster; Indicates the data point of electrical parameter change The average distance to all electrical parameter change data points in the nearest neighbor cluster; Represents the silhouette coefficient.

6. The method for monitoring the state of electric power equipment based on biosensor technology according to claim 5 is characterized in that: In S2.42, clustering the electrical parameter change data by using the K-means algorithm includes the following steps: S2.

421. For each electrical parameter change data, find the centroid closest to it and assign it to the corresponding cluster; S2.

422. For each cluster, calculate the new centroid; S2.

423. Repeat the steps until the center of mass no longer changes.

7. The method for monitoring the state of electric power equipment based on biosensor technology according to claim 6 is characterized in that: In S3, based on the analysis result, the state of the power equipment is evaluated to determine whether there is a potential fault, including the following steps: S3.

1. According to the K-means algorithm analysis results in S2, a first cluster and a second cluster are obtained; the first cluster represents a normal operating mode of low resistance and stable current, and the second cluster represents a potential fault mode of high resistance and unstable current; S3.

2. Set a threshold m for the first cluster and a threshold n for the second cluster. The threshold is used to determine the normal operating state and the potential fault state. If the currently monitored data point is close to the centroid of the first cluster, the device is in a normal state; if the data point is close to the centroid of the second cluster, the device has a potential fault.

Images