Power distribution network equipment fire risk intelligent assessment method and system based on RBI technology

Abstract

The invention discloses a power distribution network equipment fire risk intelligent assessment method and system based on the RBI technology, and belongs to the technical field of power equipment assessment, and the method comprises the steps: arranging sensors in power distribution network equipment and an operation environment thereof to collect original data; the collected original data is subjected to digital preprocessing, noise and abnormal values are removed, and processed data are output; extracting a temperature rise value, a load change value and a voltage fluctuation value from the processed data, and generating a three-dimensional health state vector reflecting an equipment operation state, an aging degree and a potential fault trend in combination with an equipment type and a historical fault record; and inputting the health state vector into a risk assessment model to generate an emergency response strategy including load adjustment, load isolation and temperature monitoring. According to the risk assessment model, the change rate of the equipment state and the offset condition of the phase space position are considered, so that the current operation state and potential risk of the equipment can be reflected more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment evaluation, and in particular to a method and system for intelligent fire risk evaluation of distribution network equipment based on RBI technology. Background Art

[0002] With the increasing complexity of distribution networks, equipment health management and fault prediction have become critical issues in the power industry. Traditional distribution network equipment maintenance and management relies primarily on regular inspections, manual patrols, and empirical judgment. However, these methods fail to fully reflect the real-time health status of equipment in actual operation and often fail to identify potential fault risks at an early stage. Therefore, the use of intelligent assessment methods based on equipment health status for real-time monitoring and risk assessment has become an important way to improve the reliability of distribution network equipment and reduce operation and maintenance costs.

[0003] Existing equipment health management technologies typically rely on a single data source or static models, making it difficult for their assessment results to fully and accurately reflect the actual status of the equipment, especially when the equipment is operating in a complex environment. While traditional risk assessment methods can predict certain faults based on historical data, their accuracy and timeliness are often limited. Existing health assessment methods often fail to consider the combined impact of multiple factors such as equipment aging, load fluctuations, and environmental changes, and are unable to provide timely risk predictions, which in turn affects the overall management and emergency response strategies of distribution network equipment.

[0004] Furthermore, while some data-analysis-based equipment health assessment methods exist, these rely on complex models and involve cumbersome calculations, making it difficult to quickly obtain equipment health status information in practical applications. Existing methods typically separate the equipment health assessment process from the fault diagnosis process, which not only increases the time delay for assessment but also complicates equipment management. Summary of the Invention

[0005] To solve the above technical problems, an intelligent fire risk assessment method for distribution network equipment based on RBI technology is proposed, including: arranging sensors in distribution network equipment and its operating environment to collect raw data; digitally preprocessing the collected raw data to remove noise and outliers, and outputting processed data; extracting temperature rise values, load change values, and voltage fluctuation values from the processed data, and combining the equipment type and historical fault records to generate a three-dimensional health status vector reflecting the equipment's operating status, aging degree, and potential failure trends; constructing a risk assessment model, inputting the health status vector into the risk assessment model, and converting the output risk assessment results into equipment fire risk levels based on a predetermined mapping relationship, and generating an emergency response strategy including load adjustment, load isolation, and temperature monitoring.

[0006] As a preferred solution of the intelligent fire risk assessment method for distribution network equipment based on RBI technology described in the present invention, the generation of a three-dimensional health status vector reflecting the equipment's operating status, aging degree and potential fault trend includes: using the first-order difference calculation method to obtain the temperature change value, load change value and voltage fluctuation value respectively on the processed data; assigning preset weights to the temperature change value, load change value and voltage fluctuation value according to the equipment type and historical fault records; synthesizing the temperature change value, load change value and voltage fluctuation value to calculate the three-dimensional health status vector, wherein the first component represents the equipment operating status, reflecting the health of the current equipment working status; the second component represents the equipment aging degree, reflecting the performance degradation of the equipment due to time and load during operation; the third component represents the potential fault trend, and assesses the risk of possible failure of the equipment in the future; the three-dimensional health status vector is recorded in the form of floating point numbers, reflecting the health status of the equipment in different dimensions.

[0007] As a preferred solution of the intelligent fire risk assessment method for distribution network equipment based on RBI technology described in the present invention, the construction of the risk assessment model includes: inputting the three-dimensional health state vector into the clustering algorithm, and the clustering algorithm classifies the health level of the equipment by calculating the similarity of the equipment characteristics according to the components of the health state vector; using a preset mapping function to map the health level classification results to a high-dimensional phase space, and the mapped equipment state reflects the health status of the equipment in the phase space; by calculating the Lyapunov exponent, the state stability of the equipment in the high-dimensional phase space is evaluated, and the risk assessment result is output.

[0008] As a preferred solution of the intelligent fire risk assessment method for distribution network equipment based on RBI technology described in the present invention, the method includes: inputting the three-dimensional health state vector of the device into the clustering algorithm, inputting the three-dimensional health state vector of the device into the K-means clustering algorithm, calculating the similarity between the components of the health state vector of the device, dividing the device into several health state groups based on the similarity, calculating the cluster center of each health state group, and outputting a clustering result including the device membership and the cluster center of each health state group; based on the clustering result, weighting each component of the health state vector of the device using the membership to obtain a weighted health state vector, and the weighted health state vector provides an optimized input for subsequent fuzzy classification and cluster analysis; using the fuzzy C-means clustering algorithm to perform fuzzy classification on the weighted health state vector, calculating the device membership in multiple health state groups, and generating a fuzzy classification result of the device; using the Mahalanobis distance to measure the similarity between the fuzzy classification result generated by the device and the cluster center of each health state group, calculating the distance between the device and the cluster center, and when the distance between the device and the cluster center meets a preset threshold, determining that the device belongs to the health state group.

[0009] As a preferred solution of the intelligent fire risk assessment method for distribution network equipment based on RBI technology described in the present invention, wherein: the clustering health status vector is mapped to a high-dimensional phase space, including using a preset mapping function to combine the device's membership in the health status group with the health status vector, and calculating the mapping coefficient to determine the device's position in the high-dimensional phase space; using a multidimensional function mapping algorithm, using the mapping coefficient as input, mapping the device health status vector to the high-dimensional phase space, and generating the device's phase space coordinates; determining the device's position in the phase space based on the device's phase space coordinates; updating the device's health status distribution based on the device's position in the phase space, and adjusting the device's health status assessment based on the position change in the phase space.

[0010] As a preferred solution of the intelligent fire risk assessment method for distribution network equipment based on RBI technology described in the present invention, wherein: the calculation of the Lyapunov exponent includes collecting state data of the equipment in phase space, obtaining the state vector of the equipment at multiple time points, and recording the state information of the equipment at each discrete time point; for any two adjacent time points, calculating the change in the equipment state vector to obtain the equipment state change rate;

[0011] Based on the state change rates at multiple time points, the average rate of device state change is calculated, and based on the average rate of device state change, the Lyapunov exponent is calculated.

[0012] As a preferred solution of the intelligent fire risk assessment method for distribution network equipment based on RBI technology described in the present invention, the output risk assessment results include: when the Lyapunov index is greater than a preset threshold and the position of the equipment state in the phase space is offset, the equipment is judged to be a first-level risk equipment, and emergency response measures are immediately implemented; when the Lyapunov index is greater than the preset threshold and the position of the equipment state in the phase space is not offset, the equipment is judged to be a second-level risk equipment, and conventional monitoring measures are implemented; when the Lyapunov index is less than or equal to the preset threshold and the position of the equipment state in the phase space is offset, the equipment is judged to be a third-level risk equipment, and load adjustment strategy and temperature monitoring strategy are implemented; when the Lyapunov index is less than or equal to the preset threshold and the position of the equipment state in the phase space is not offset, the equipment is judged to be a fourth-level risk equipment, and only regular inspection and conventional monitoring are implemented.

[0013] Another objective of the present invention is to provide an intelligent fire risk assessment system for distribution network equipment based on RBI technology. This system addresses the issues of poor real-time performance and incomplete assessment in existing distribution network equipment health management. Traditional methods fail to effectively combine historical and real-time data, making it difficult to predict and respond to equipment failures in a timely manner. Furthermore, existing technologies ignore the combined impact of equipment aging and environmental factors, are unable to dynamically adjust equipment status assessments and risk levels, and lack an effective multi-dimensional risk assessment mechanism.

[0014] As a preferred solution of the intelligent fire risk assessment system for distribution network equipment based on RBI technology described in the present invention, it is characterized by including: a data acquisition module, which is used to arrange sensors in the distribution network equipment and its operating environment to collect raw data; a preprocessing module, which is used to digitally preprocess the collected raw data, eliminate noise and outliers, and output processed data; a feature extraction module, which is used to extract temperature rise values, load change values and voltage fluctuation values from the processed data, and combine the equipment type and historical fault records to generate a three-dimensional health status vector reflecting the equipment's operating status, aging degree and potential fault trend; and a risk assessment module, which is used to construct a risk assessment model, input the health status vector into the risk assessment model, convert the output risk assessment result into an equipment fire risk level according to a predetermined mapping relationship, and generate an emergency response strategy including load adjustment, load isolation and temperature monitoring.

[0015] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the intelligent fire risk assessment method for distribution network equipment based on RBI technology are implemented.

[0016] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the intelligent fire risk assessment method for distribution network equipment based on RBI technology.

[0017] The present invention's beneficial effects: By combining the device's health state vector with its high-dimensional phase space coordinates and employing the Lyapunov exponent to assess device stability, the system can effectively improve the accuracy of device failure prediction and risk identification. Using clustering algorithms and fuzzy classification techniques, the system accurately classifies and assesses device health, minimizing human interference and automatically generating device risk scores.

[0018] By considering the rate of change of device states and the offset of their phase space positions, the risk assessment model of this invention can more accurately reflect the current operating status and potential risks of devices, ensuring that devices receive appropriate response strategies at different risk levels, further enhancing the safety and reliability of distribution network equipment. In particular, during emergency responses for high-risk equipment, timely load adjustment and temperature monitoring strategies can effectively reduce the probability of fire incidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 This is a general flow chart of a method for intelligent fire risk assessment of distribution network equipment based on RBI technology provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0021] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0022] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a method for intelligently assessing fire risk of distribution network equipment based on RBI technology, including:

[0023] Step 1: Deploy sensors in distribution network equipment and their operating environment to collect raw data;

[0024] It should be noted that temperature sensors are arranged to monitor the temperature rise of the equipment in real time. Especially under high-load operation, the temperature sensors can measure the temperature changes of the equipment and promptly reflect whether the equipment is overloaded or whether there is a potential failure risk.

[0025] Current and voltage sensors are deployed to collect load and voltage data. Current sensors can monitor changes in load current in real time, while voltage sensors can monitor voltage fluctuations. Load data is obtained by multiplying the current and voltage (power).

[0026] For equipment such as transformers, ensure that the sensor is installed in the core part of the equipment, such as the winding, oil tank or cooling system, to ensure that the temperature rise data reflects the actual operating status of the equipment.

[0027] Step 2: The collected raw data is digitally preprocessed to remove noise and outliers, and the processed data is output;

[0028] It should be noted that the collected raw data may be affected by external interference or the device's inherent characteristics. Therefore, preprocessing operations such as filtering and denoising are required. Preprocessing is particularly critical for data on temperature rise, load variation, and voltage fluctuation to ensure data accuracy and reliability. Common preprocessing methods include smoothing, filtering algorithms (such as low-pass filters), and outlier detection. Preprocessed data can more accurately reflect the actual health status of the device.

[0029] Furthermore, algorithms such as the sliding average method are used to smooth the data to reduce data fluctuations caused by measurement errors or external interference. The length of the sliding window should be adjusted according to the sampling frequency and the rate of change of the equipment.

[0030] Using digital filtering algorithms (such as Kalman filtering or low-pass filtering) to denoise temperature rise, load, and voltage data can effectively improve data quality, especially in environments with high-frequency oscillations or high noise levels of power grid equipment.

[0031] Detect and remove outliers based on statistical methods (such as standard deviation or IQR). Outliers are often caused by equipment failure, sensor failure, or environmental events. Removing outliers ensures the accuracy of subsequent data analysis.

[0032] Step 3: Extract temperature rise values, load change values, and voltage fluctuation values from the processed data. Combined with the equipment type and historical fault records, a three-dimensional health status vector is generated that reflects the equipment's operating status, aging level, and potential fault trends.

[0033] In step 3, generating a three-dimensional health state vector reflecting the equipment's operating status, aging degree, and potential failure trend includes using a first-order difference calculation method on the processed data to obtain temperature change values, load change values, and voltage fluctuation values respectively;

[0034] According to the equipment type and historical fault records, the temperature change value, load change value and voltage fluctuation value are assigned preset weights respectively;

[0035] The temperature change, load change, and voltage fluctuation values are combined to calculate a three-dimensional health status vector. The first component represents the equipment's operating status, reflecting the current health of the equipment. The second component represents the equipment's aging, reflecting the performance degradation caused by time and load during operation. The third component represents the potential failure trend, assessing the risk of future equipment failure.

[0036] The three-dimensional health status vector is recorded in the form of floating-point numbers, reflecting the health status of the device in different dimensions.

[0037] It should be noted that the first-order difference method calculates the change in a variable by calculating the difference between data points at adjacent time points. For temperature, load, and voltage, the results of the first-order difference can reflect the changing trend of the equipment status within each time period.

[0038] The temperature change value is calculated by setting the time series T t Indicates the temperature value at time point t. According to the first-order difference calculation method, the temperature change value ΔT t It can be expressed as:

[0039] ΔT t =T t -T t-1

[0040] Among them, T t is the temperature at the current time, T t-1 is the temperature at the previous point in time.

[0041] The load change value is calculated by the load change value ΔP t It is also calculated as the difference in loads at adjacent time points, using the following formula:

[0042] ΔP t =P t -P t-1

[0043] Among them, P t and P t-1 are the load values at the current time point and the previous time point respectively.

[0044] The voltage fluctuation value is calculated by the voltage fluctuation value ΔV t The calculation is the same as for the previous two values:

[0045] ΔV t =V t -V t-1

[0046] Among them, V t and V t-1 are the voltage values at the current and previous time points respectively.

[0047] Through first-order difference calculation, the changes in temperature, load and voltage in each time interval are obtained, reflecting the changes in the operating status of the equipment at different time points.

[0048] Furthermore, based on the equipment type and historical fault records, preset weights will be assigned to temperature change values, load change values, and voltage fluctuation values. The way the weights are assigned is closely related to the characteristics of the equipment. For example, for high-temperature equipment, the weight of the temperature change value will be set to be higher to ensure that this data dominates when assessing the health status of the equipment. For load-sensitive equipment (such as transformers or generators), the load change value will be given a higher weight, which can more accurately reflect whether the equipment is overloaded and the impact on equipment aging.

[0049] At the same time, historical failure records also influence how weights are assigned. If a device's history indicates multiple failures due to overheating or overload, these failure records will increase the weight of temperature and load-related data. For example, for a device that has experienced two or more overheating failures, the weight of the temperature change value will be significantly increased, which will enhance the model's ability to respond to potential overheating issues. The device's failure records will be dynamically weighted using machine learning algorithms or rule engines to ensure that historical failure trends are properly reflected in the model's output.

[0050] On this basis, the temperature change value, load change value and voltage fluctuation value will be weighted and synthesized to generate a three-dimensional health status vector. This vector contains three main components:

[0051] The first component represents the device's operating status, reflecting its current health. This component is calculated as a weighted average of temperature change, load change, and voltage fluctuation. Specifically, if the device's temperature continues to rise, the load approaches or exceeds the maximum load capacity, and the voltage fluctuates significantly, these factors will be weighted and combined to form the basis for evaluating the device's current operating status.

[0052] The second component represents the degree of equipment aging, assessing the performance degradation of equipment due to load and environmental factors over long periods of operation. This component is calculated using historical load data, the equipment's service life, and long-term temperature fluctuation data. For example, operating equipment at higher temperatures accelerates the aging process, and historical temperature fluctuation data will be weighted into this component, affecting the aging assessment.

[0053] The third component represents potential failure trends, assessing the risk of future equipment failures. This component combines the combined effects of temperature changes, load changes, and voltage fluctuations. Excessive temperature or overload can lead to unstable operation, increasing the risk of potential failure. The system automatically estimates the potential failure risk of equipment based on historical data, current status, and the equipment's operating mode.

[0054] Specifically, temperature fluctuations often directly impact the degree of device aging, particularly at higher temperatures, which can accelerate aging. Increased loads can also cause device temperatures to rise, accelerating degradation. Therefore, temperature changes should be closely correlated with the degree of device aging.

[0055] Based on the equipment's historical failure records, the impact of temperature changes on equipment aging can be adjusted using a weighting factor. Temperature changes under different load conditions accelerate aging, so a higher weight should be assigned to temperature changes.

[0056] Load changes are directly related to equipment failure trends. High loads or frequently fluctuating loads increase the risk of equipment failure, especially under high loads, which can lead to equipment overloads, failures, and other problems. Therefore, load changes should be linked to potential failure trends.

[0057] Based on the equipment's usage history and statistical analysis of load fluctuations, a data mining model (e.g., K-means clustering or fuzzy C-means) is used to calculate a weight associated with failure trends for each load fluctuation. If a device is frequently under high load, its failure risk weight should be higher.

[0058] Voltage fluctuations will directly affect the working stability of equipment, especially in distribution networks. Voltage instability may cause frequent short-term failures or long-term performance degradation of equipment.

[0059] Based on the voltage variation and its impact on equipment stability, we can assign a moderate weight to voltage fluctuation, especially when voltage fluctuations are frequent and the equipment has experienced voltage-related faults, giving voltage variation a higher weight.

[0060] According to the dynamically adjusted weight (W T ,W L ,W V ), weighted synthesis of temperature change, load change and voltage fluctuation is performed to calculate the three-dimensional health state vector, which is expressed as,

[0061] H1=W T ΔT,H2=W L ΔL,H3=W V ΔV

[0062] Among them, H1 represents the equipment operating status, indicating the current health of the equipment; H2 represents the equipment aging degree, reflecting the performance degradation of the equipment due to load and temperature changes; H3 represents the potential failure trend, assessing the risk of future equipment failure.

[0063] The synthesized health state vector:

[0064] H=[H1,H2,H3]

[0065] This three-dimensional vector will serve as a health assessment indicator for the device, reflecting the overall health status of the device.

[0066] It should be noted that by collecting multi-dimensional data such as the temperature rise, load change, and voltage fluctuation of the equipment, and combining it with the equipment type and historical fault records, a three-dimensional health status vector is generated. This method breaks through the limitations of traditional single indicator evaluation and no longer relies solely on a single parameter (such as temperature or load fluctuation) to evaluate the health status of the equipment. Temperature fluctuations, load changes, and voltage fluctuations are key factors reflecting the status of the equipment, but there is often an interaction between them. For example, load fluctuations may cause the temperature of the equipment to rise, and temperature changes may affect the electrical performance of the equipment. Using each dimension alone may lead to misjudgment of the health status of the equipment. Therefore, the present invention makes the evaluation results more accurate and reliable by considering the interaction between the various parameters.

[0067] Traditional technologies typically use fixed weight coefficients to simply perform weighted calculations on various indicators. However, during equipment operation, the impact of load and temperature changes on the equipment is nonlinear, and the extent of these impacts varies with the equipment's usage history and operating environment. Therefore, a dynamic weight adjustment mechanism can assign appropriate weights to different data dimensions based on the equipment's actual operating status and historical fault records, accurately reflecting the equipment's current status at each moment. For example, for a device that frequently experiences overload failures, its load changes may have a greater impact on health assessments, while for a severely aged device, the impact of temperature changes will be more prominent. Dynamically adjusted weights make equipment health assessments more consistent with the equipment's actual operating conditions, improving the accuracy and adaptability of the assessment.

[0068] Traditional health assessments often rely on simple classification methods, such as categorizing devices into broad categories like "good," "fair," and "faulty," which fail to provide in-depth analysis of a device's specific health status. In contrast, the present invention uses a clustering algorithm to input device health status vectors, meticulously categorizing devices into distinct health status groups and calculating the cluster centers for each group, thereby accurately classifying devices into more detailed categories. This process allows for precise device classification based on their various status characteristics, avoiding the "oversimplification" of traditional methods and providing more accurate health assessment results.

[0069] After highly accurate classification of health status, high-dimensional phase space mapping technology is employed to map the health status vector into high-dimensional phase space, providing a solid data foundation for analyzing equipment stability and predicting future risks. By calculating the Lyapunov exponent, the stability of the equipment in phase space can be assessed in real time, particularly when the equipment faces potential failure risks, enabling timely identification of stability trends. Traditional equipment monitoring methods often struggle to effectively predict the future state of an equipment, but the Lyapunov exponent provides valuable early warning information to equipment managers by monitoring the rate of change of equipment status in real time.

[0070] Combining high-dimensional mapping with Lyapunov exponent analysis enables multi-level risk assessment of equipment. Based on the equipment's risk score, the system determines its risk level and adopts an appropriate emergency response strategy based on the equipment's risk level. Unlike traditional solutions, this approach dynamically adjusts response strategies based on the specific state of the equipment. For example, it implements timely load adjustments and temperature monitoring in the event of equipment status deviations, thereby minimizing the risk of failure and equipment loss.

[0071] Step 4: Build a risk assessment model, input the health status vector into the risk assessment model, convert the output risk assessment results into equipment fire risk levels according to a predetermined mapping relationship, and generate an emergency response strategy including load adjustment, load isolation and temperature monitoring.

[0072] In step 4, building the risk assessment model includes inputting the three-dimensional health state vector into a clustering algorithm. The clustering algorithm classifies the health level of the equipment by calculating the similarity of the equipment characteristics based on each component of the health state vector;

[0073] Use a preset mapping function to map the health classification results to a high-dimensional phase space. The mapped device status reflects the health status of the device in the phase space.

[0074] By calculating the Lyapunov exponent, the state stability of the equipment in the high-dimensional phase space is evaluated and the risk assessment results are output.

[0075] It should be noted that unlike traditional methods that use simple threshold judgments or single-factor analysis, this method uses a comprehensive assessment based on a multidimensional health status vector. This allows for a comprehensive assessment of the device's health from multiple perspectives, significantly improving the accuracy and reliability of the assessment. Furthermore, a clustering algorithm automatically identifies device health status categories based on their historical status. This allows the entire assessment process to not only reflect the device's current health level in real time but also predict future failure risks based on the device's historical operating conditions.

[0076] The advantage of this approach is that the clustering algorithm automatically adjusts the classification criteria based on the similarity of device characteristics, rather than relying on fixed thresholds set by humans. This avoids the static assessment issues that can exist with traditional methods. This technology makes device health assessment more flexible and accurate, providing a more precise basis for risk management.

[0077] Inputting the three-dimensional health state vector into the clustering algorithm includes inputting the three-dimensional health state vector of the device into a K-means clustering algorithm, calculating similarities between components of the health state vector of the device, dividing the devices into a number of health state groups based on the similarities, calculating a cluster center of each health state group, and outputting a clustering result including a membership degree of the device and a cluster center of each health state group;

[0078] According to the clustering results, the membership degree is used to weight each component of the equipment health status vector to obtain a weighted health status vector. The weighted health status vector provides an optimized input for subsequent fuzzy classification and clustering analysis.

[0079] Use the fuzzy C-means clustering algorithm to perform fuzzy classification on the weighted health status vector, calculate the device's membership in multiple health status groups, and generate the device's fuzzy classification results;

[0080] The Mahalanobis distance is used to measure the similarity between the fuzzy classification results generated by the device and the cluster center of each health status group. The distance between the device and the cluster center is calculated. When the distance between the device and the cluster center meets the preset threshold, the device is judged to belong to the health status group.

[0081] It should be noted that traditional equipment health assessments often rely on thresholds or data from a single variable. These methods often overlook the complexity of a device's multi-dimensional health status and fail to accurately reflect the device's true condition in complex operating environments. Consequently, using a single indicator or static assessment criteria can lead to biased assessment results and may even miss early warning signs of equipment failure.

[0082] This embodiment introduces a health status classification method based on the K-means clustering algorithm. This clustering algorithm calculates the similarity between the components of the health status vector and divides devices into different health status groups. The innovation of this technical solution lies in the dynamic classification of health status based on the multi-dimensional data of the device and the optimization of the health status vector through membership weighting, providing optimized input for subsequent fuzzy classification and cluster analysis. This has significant advantages over existing health assessment methods that rely on fixed rules.

[0083] By employing a clustering algorithm, classification criteria can be dynamically adjusted based on the actual operating status of the device, making device health assessment more flexible and accurate. The clustering results provide a data-driven "health label" for each device, avoiding the human-generated bias of traditional methods and achieving greater accuracy and flexibility. Compared to existing technologies, this approach can better adapt to the operating status of different devices and provide precise assessments of device health under different operating conditions.

[0084] In an optional embodiment of the present invention, the three-dimensional health status vector of the device is first input into a K-means clustering algorithm. The purpose of the K-means algorithm is to group the health status of the device by minimizing the distance between the device and the cluster center.

[0085] Step 1: Calculate the Euclidean distance between the device and the cluster center:

[0086] Given a three-dimensional health state vector H of a device i ={H i1 ,H i2 ,H i3} and cluster center C k ={C k1 ,C k2 ,C k3}, first calculate the Euclidean distance between the device and each cluster center:

[0087]

[0088] Among them, d ik Represents the distance between device i and cluster center k. This distance metric is used to measure the similarity between the device and the cluster center.

[0089] Step 2: Assign the device to the closest cluster center:

[0090] According to the calculated distance d ik , assign each device i to the cluster center C closest to it k ,Right now:

[0091]

[0092] Device i is assigned to the minimum distance d ik The corresponding cluster center C k .

[0093] Step 3: Update the cluster centers:

[0094] After all devices are assigned, the cluster center C k It needs to be updated based on the mean of the device health status vector. The new cluster center Determined by the average value of the health status vectors of the devices belonging to the cluster:

[0095]

[0096] Among them, N k is the number of devices in cluster k. In this way, we update the location of the cluster center, which is determined by averaging the health status vectors of all devices.

[0097] Step 4: Iteratively update the cluster centers:

[0098] Steps 2 and 3 will be iterated continuously until the position change of the cluster center is less than the preset threshold and the clustering results converge.

[0099] Calculate membership and weight it: After obtaining the K-means clustering results, we use the distance d between the device and the cluster center to ik Calculate the device's membership. Membership μ ik Indicates that device i belongs to cluster center C k The degree of health status of the device is calculated based on the distance between the device and the cluster center. In order to prevent the weight of the cluster center from being too large or too small, we introduce a weighted method to weight the health status vector of the device.

[0100] Membership degree μ ik It can be calculated by the following formula:

[0101]

[0102] in, is the minimum distance between device i and all cluster centers. This formula indicates that the smaller the distance between device i and cluster center k, the greater the degree of membership, and the higher the degree of membership of the device to the cluster center.

[0103] According to the calculated membership, the three-dimensional health state vector of each device is weighted. The weighted health state vector Calculated by the following formula:

[0104]

[0105] Among them, H i is the original health status vector of device i, μ ik is the degree of membership of device i to cluster center k, is the weighted health state vector.

[0106] After obtaining the weighted health state vector, the fuzzy C-means clustering algorithm is used to classify the devices. The fuzzy C-means clustering algorithm classifies devices by membership, and a device can belong to multiple clusters at the same time. We use the following formula to minimize the objective function J of the fuzzy C-means clustering m :

[0107]

[0108] Among them, μ ik is the degree of membership of device i to cluster center k, m is the fuzzy index, usually 2,d ik is the Euclidean distance between device i and cluster center k. Minimize J m A balance is achieved between the device's membership and distance, so that the device can be accurately classified.

[0109] The Mahalanobis distance is used to measure the similarity between the device and each cluster center. The Mahalanobis distance takes into account the correlation between the dimensions of the health status vector and is calculated using the following formula:

[0110]

[0111] Among them, H i is the health status vector of device i, C k is the health status vector of cluster center k, S -1 is the inverse of the covariance matrix of the health state vector. The use of Mahalanobis distance can effectively avoid the limitations of Euclidean distance, especially when there is correlation between different dimensions of the health state vector.

[0112] According to the calculated Mahalanobis distance d M , to determine whether the device belongs to a certain health status group. If the distance d between the device and the cluster center is M If the value is less than the preset threshold, the device is considered to belong to this health status group:

[0113] if d M ≤threshold then Device i belongs to C k

[0114] If the distance between the device and the cluster center is less than a threshold, the device is determined to belong to the health status group.

[0115] Mapping the clustered health state vector to the high-dimensional phase space includes combining the device's membership in the health state group with the health state vector using a preset mapping function, and calculating a mapping coefficient to determine the device's position in the high-dimensional phase space;

[0116] Using a multidimensional function mapping algorithm, the mapping coefficient is used as input to map the device health state vector to a high-dimensional phase space and generate the phase space coordinates of the device;

[0117] According to the phase space coordinates of the device, the position of the device in the phase space is determined. The position reflects the health of the device. A position close to the center of the phase space indicates that the device is in good health, while a position away from the center indicates that the device is in poor health.

[0118] Based on the position of the device in the phase space, the health status distribution of the device is updated, and the health status assessment of the device is adjusted according to the position change in the phase space.

[0119] It should be noted that traditional equipment health assessment methods typically rely solely on static assessment of equipment status through basic monitoring data (such as temperature, load, and voltage). However, these traditional methods fail to fully consider the temporal changes in equipment status and its dynamic response in complex environments. Equipment health status reflects not only instantaneous values but also a comprehensive reflection of long-term operation, environmental changes, and load fluctuations.

[0120] This embodiment introduces high-dimensional phase space mapping. By mapping the device's health status vector into high-dimensional space, it accurately captures the device's temporal changes and interrelationships. The innovation of this technical solution lies in its use of a multidimensional function mapping algorithm to convert the device's health status vector into coordinates in high-dimensional space, which then reflect the device's evolving health status. This method not only displays the device's health status but also captures potential changes in the device's future operation.

[0121] Compared to conventional static assessment methods, the high-dimensional phase space mapping technology of the present invention can capture dynamic changes in device health status in real time, providing a more intuitive and accurate health assessment method. By combining device status with position changes in high-dimensional phase space, it is possible to effectively identify device operational anomalies and promptly adjust risk assessment strategies. This method is particularly suitable for complex systems such as distribution network equipment, dynamically responding to load changes and external environmental fluctuations, and improving the accuracy of device health assessments.

[0122] In an optional embodiment of the present invention, the membership information and the health state vector are combined using a weighted average method to obtain the mapping coefficient of the device position in the phase space, which is expressed as:

[0123]

[0124] Among them, μ ik is the cluster center C of device i k The membership degree of is the health status vector of device i in cluster k. The mapping coefficient γ is calculated by weighting the health status vector of the corresponding cluster center. i , which represents the preliminary mapping position of the device in the high-dimensional phase space.

[0125] The mapping coefficient γ i As input, a multidimensional function mapping algorithm is used to map the health state vector of the device to a high-dimensional phase space. This step can be achieved by a nonlinear mapping function, which can be a transformation in the high-dimensional space, such as a polynomial mapping or a Gaussian mapping. The mapping formula used is:

[0126] S i =f(γ i )

[0127] Among them, S i is the coordinate of device i in the high-dimensional phase space, f(γ i ) is a multidimensional function mapping algorithm that maps the device's health status vector to a new coordinate space. The function f(·) here can use different mapping algorithms, such as Gaussian kernel function, radial basis function, etc., depending on the actual application.

[0128] The coordinate S of device i in the high-dimensional phase space is obtained through the multi-dimensional mapping algorithm i Next, the health of the device is assessed based on its position in phase space.

[0129] The position of a device reflects its health: a position close to the center of the phase space indicates good health, while a position away from the center indicates poor health. The center of the phase space represents the ideal health state, and the degree of deviation from the center represents the difference in health, expressed as,

[0130] d i =||S i -S center ||

[0131] Among them, d i is the distance between device i and the phase space center S in the high-dimensional phase space center The distance between them, ∥·∥ represents the Euclidean distance. If the device position S i A closer distance to the center indicates a good device health status; conversely, a larger position offset indicates a poor device health status.

[0132] Based on the position d of the device in the high-dimensional phase spacei , update the health status distribution of the device. This process is dynamic, and the health status evaluation of the device will be adjusted as the position of the device in the phase space changes, which can be expressed as,

[0133]

[0134] in, is the updated health status vector, is the current health state vector, λ is the adjustment coefficient, d i is the distance between device i and the center of the phase space. By adjusting the coefficient λ, the health status of the device will change according to its position in the phase space, thus achieving dynamic adjustment of the health status assessment.

[0135] Calculating the Lyapunov exponent includes collecting the state data of the device in the phase space, obtaining the state vector of the device at multiple time points, and recording the state information of the device at each discrete time point;

[0136] For any two adjacent time points, calculate the change of the device state vector and obtain the device state change rate;

[0137] Based on the state change rates at multiple time points, the average rate of device state change is calculated, and based on the average rate of device state change, the Lyapunov exponent is calculated.

[0138] It should be noted that traditional equipment health assessments typically rely on simple status values or operating parameters, but these methods often fail to provide in-depth insights into equipment stability or future risks. For power distribution systems, the health of equipment not only reflects its current state but is also closely linked to future failures. Therefore, static assessments based on simple data cannot meet the risk management needs of complex environments.

[0139] This embodiment method uses the Lyapunov exponent to assess the dynamic stability of a device in high-dimensional phase space. As an important tool for measuring system stability, the Lyapunov exponent can effectively reveal the instability of a device during dynamic changes. By recording the trajectory of the device state in phase space and calculating the rate of state change at adjacent time points, the present invention can promptly identify whether a device is approaching an unstable state.

[0140] This technical solution is innovative compared to traditional assessment methods in that it predicts potential equipment failures in advance by dynamically analyzing the equipment's operational stability. The Lyapunov index provides a quantitative measure of equipment health, reflecting whether it is critical, thus providing a basis for timely preventive measures. While traditional methods often rely on a single monitoring indicator, the Lyapunov index of this invention provides a new risk assessment method based on the dynamic changes in equipment status, significantly improving the accuracy and preemptiveness of risk predictions.

[0141] In an optional embodiment of the present invention, the state data of the device in the phase space is collected, and the state vectors S(t1), S(t2), ..., S(t n ), where S(t i ) indicates that the device is at time t i The state vector at each moment. These state vectors represent the health status of the device at different moments, and the formula is:

[0142] S(t i )={H1(t i ),H2(t i ),H3(t i )},i=1,2,…,n

[0143] Among them, H1(t i ),H2(t i ),H3(t i ) are the devices at time t i The three components of the health status vector at a certain moment correspond to the operating status, aging degree and potential failure trend of the equipment.

[0144] For any two adjacent time points t i and t i+1 , calculate the change in the device state vector ΔS i =S(t i+1 )-S(t i ), and calculate the rate of change of the device state based on this change The formula is:

[0145] ΔS i =S(t i+1 )-S(t i )

[0146]

[0147] Where Δt = t i+1 -t i Represents the time interval between two time points. Indicates the rate of change of the device status, that is, the change in the device health status per unit time.

[0148] Next, based on the state change rate at multiple time points We calculate the average rate of device state change. By averaging multiple state change rates, we can more accurately reflect the health change trend of the device over a period of time. The formula is:

[0149]

[0150] in, is the average rate of device state change, and n is the total number of sampled time points.

[0151] Based on the average rate of device state changes Calculate the Lyapunov exponent. The Lyapunov exponent is used to evaluate the stability of the device in phase space and indicates the sensitivity of the system to initial conditions. If the Lyapunov exponent is positive, the device state is unstable, otherwise it is stable. The formula is:

[0152]

[0153] Where λ is the Lyapunov exponent. This exponent measures the device's sensitivity to its initial state after multiple state changes. A larger value indicates a more unstable system, while a smaller value indicates a more stable system. This formula calculates the logarithm of each state change to determine the stability of the device over time in phase space.

[0154] The output risk assessment results include: when the Lyapunov exponent is greater than the preset threshold and the position of the device state in the phase space is offset, the device is judged to be a level 1 risk device and emergency response measures are immediately implemented;

[0155] When the Lyapunov exponent is greater than the preset threshold and the position of the device state in the phase space has not shifted, the device is judged to be a level 2 risk device and conventional monitoring measures are implemented;

[0156] When the Lyapunov exponent is less than or equal to the preset threshold and the position of the equipment state in the phase space is offset, the equipment is judged to be a level 3 risk equipment and the load adjustment strategy and temperature monitoring strategy are implemented;

[0157] When the Lyapunov exponent is less than or equal to the preset threshold and the position of the device state in the phase space has not shifted, the device is judged to be a level 4 risk device and only periodic inspections and routine monitoring are performed.

[0158] Example 2 is the second embodiment of the present invention, which is different from the previous embodiment in that:

[0159] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0160] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0161] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0162] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0163] Example 3, the third embodiment of the present invention, provides an intelligent fire risk assessment system for distribution network equipment based on RBI technology, including.

[0164] A data acquisition module is used to arrange sensors in the distribution network equipment and its operating environment to collect raw data;

[0165] The preprocessing module is used to preprocess the collected raw data, remove noise and outliers, and output the processed data;

[0166] A feature extraction module is used to extract temperature rise values, load change values, and voltage fluctuation values from the processed data, and combine the equipment type and historical fault records to generate a three-dimensional health status vector that reflects the equipment's operating status, aging level, and potential fault trends; and

[0167] The risk assessment module is used to build a risk assessment model, input the health status vector into the risk assessment model, convert the output risk assessment results into equipment fire risk levels according to a predetermined mapping relationship, and generate emergency response strategies including load adjustment, load isolation and temperature monitoring.

[0168] It should be noted that the above embodiments are only used to illustrate the technical solutions 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 preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent fire risk assessment method for distribution network equipment based on RBI technology is characterized by: include, Deploy sensors on distribution network equipment and their operating environment to collect raw data; The collected raw data is digitally preprocessed to remove noise and outliers, and the processed data is output; Extract temperature rise, load change, and voltage fluctuation values from the processed data. Combined with equipment type and historical fault records, a three-dimensional health status vector is generated that reflects the equipment's operating status, aging level, and potential fault trends. Construct a risk assessment model, input the health status vector into the risk assessment model, convert the output risk assessment results into equipment fire risk levels according to a predetermined mapping relationship, and generate an emergency response strategy including load adjustment, load isolation and temperature monitoring.

2. The method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to claim 1, characterized in that: The generation of a three-dimensional health status vector reflecting the equipment's operating status, aging level, and potential failure trends includes: The first-order difference calculation method is used on the processed data to obtain the temperature change value, load change value and voltage fluctuation value respectively; According to the equipment type and historical fault records, the temperature change value, load change value and voltage fluctuation value are assigned preset weights respectively; The temperature change, load change, and voltage fluctuation values are combined to calculate a three-dimensional health status vector. The first component represents the equipment's operating status, reflecting the current health of the equipment. The second component represents the equipment's aging, reflecting the performance degradation caused by time and load during operation. The third component represents the potential failure trend, assessing the risk of future equipment failure. The three-dimensional health status vector is recorded in the form of floating-point numbers, reflecting the health status of the device in different dimensions.

3. The method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to claim 2, characterized in that: The risk assessment model construction includes: The three-dimensional health state vector is input into the clustering algorithm, which classifies the health level of the devices by calculating the similarity of the device characteristics based on the components of the health state vector; Use a preset mapping function to map the health classification results to a high-dimensional phase space. The mapped device status reflects the health status of the device in the phase space. By calculating the Lyapunov exponent, the state stability of the equipment in the high-dimensional phase space is evaluated and the risk assessment results are output.

4. The method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to claim 3, characterized in that: The inputting of the three-dimensional health state vector into the clustering algorithm includes: The three-dimensional health status vector of the device is input into the K-means clustering algorithm. The similarity between the components of the device health status vector is calculated. The devices are divided into several health status groups based on the similarity. The cluster center of each health status group is calculated. The clustering result including the device membership degree and the cluster center of each health status group is output. According to the clustering results, the membership degree is used to weight each component of the equipment health status vector to obtain a weighted health status vector. The weighted health status vector provides an optimized input for subsequent fuzzy classification and clustering analysis. Use the fuzzy C-means clustering algorithm to perform fuzzy classification on the weighted health status vector, calculate the device's membership in multiple health status groups, and generate the device's fuzzy classification results; The Mahalanobis distance is used to measure the similarity between the fuzzy classification results generated by the device and the cluster center of each health status group. The distance between the device and the cluster center is calculated. When the distance between the device and the cluster center meets the preset threshold, the device is judged to belong to the health status group.

5. The method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to claim 4, characterized in that: The clustered health state vector is mapped to the high-dimensional phase space include, Using a preset mapping function to combine the device's membership in the health state group with the health state vector, and calculate a mapping coefficient that determines the device's position in the high-dimensional phase space; Using a multidimensional function mapping algorithm, the mapping coefficient is used as input to map the device health state vector to a high-dimensional phase space and generate the phase space coordinates of the device; Determine the position of the device in the phase space according to the phase space coordinates of the device; Based on the position of the device in the phase space, the health status distribution of the device is updated, and the health status assessment of the device is adjusted according to the position change in the phase space.

6. The method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to claim 5, characterized in that: The calculation of the Lyapunov exponent includes, Collect the state data of the device in the phase space, obtain the state vector of the device at multiple time points, and record the state information of the device at each discrete time point; For any two adjacent time points, calculate the change of the device state vector and obtain the device state change rate; Based on the state change rates at multiple time points, the average rate of device state change is calculated, and based on the average rate of device state change, the Lyapunov exponent is calculated.

7. The method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to claim 6, characterized in that: The output risk assessment results include: When the Lyapunov exponent is greater than the preset threshold and the position of the device state in the phase space is offset, the device is judged to be a level 1 risk device and emergency response measures are immediately implemented; When the Lyapunov exponent is greater than the preset threshold and the position of the device state in the phase space has not shifted, the device is judged to be a level 2 risk device and conventional monitoring measures are implemented; When the Lyapunov exponent is less than or equal to the preset threshold and the position of the equipment state in the phase space is offset, the equipment is judged to be a level 3 risk equipment and the load adjustment strategy and temperature monitoring strategy are implemented; When the Lyapunov exponent is less than or equal to the preset threshold and the position of the device state in the phase space has not shifted, the device is judged to be a level 4 risk device and only periodic inspections and routine monitoring are performed.

8. A distribution network equipment fire risk intelligent assessment system based on RBI technology, applying the distribution network equipment fire risk intelligent assessment method based on RBI technology as described in any one of claims 1 to 7, characterized in that: include: A data acquisition module is used to arrange sensors in the distribution network equipment and its operating environment to collect raw data; The preprocessing module is used to preprocess the collected raw data, remove noise and outliers, and output the processed data; A feature extraction module is used to extract temperature rise values, load change values, and voltage fluctuation values from the processed data, and combine the equipment type and historical fault records to generate a three-dimensional health status vector that reflects the equipment's operating status, aging level, and potential fault trends; and The risk assessment module is used to build a risk assessment model, input the health status vector into the risk assessment model, convert the output risk assessment results into equipment fire risk levels according to a predetermined mapping relationship, and generate emergency response strategies including load adjustment, load isolation and temperature monitoring.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for intelligent fire risk assessment of distribution network equipment based on RBI technology according to any one of claims 1 to 7 are implemented.

Images