CN-121978174-A - Power station equipment state degradation early warning method

CN121978174ACN 121978174 ACN121978174 ACN 121978174ACN-121978174-A

Abstract

The invention provides a power station equipment state degradation early warning method which comprises the steps of integrating a charge sensor and a high-precision dielectric sensor in a lubricating oil circulation system of power station equipment, setting an acquisition time interval, dividing a target acquisition time period into acquisition time points according to the preset acquisition time interval, further preliminarily acquiring charge data of power station equipment oil products at the acquisition time points, identifying an oil degradation preliminary evaluation coefficient of the lubricating oil corresponding to the power station equipment based on the charge data, identifying whether the oil degradation exists in the lubricating oil circulation system corresponding to the power station equipment, synchronously extracting dielectric data of the power station equipment oil products at the acquisition time points based on the arranged high-precision dielectric sensor, deeply evaluating the oil state of the lubricating oil circulation system corresponding to the power station equipment, and sending corresponding early warning signals. Through multi-parameter fusion analysis, the monitoring precision is improved, initial signs of oil degradation can be found earlier, and equipment faults caused by oil performance degradation are avoided.

Inventors

HU XIN
LAI XINJIE
LUO YUANLIN
WU YUECHAO

Assignees

中国电建集团华东勘测设计研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20251201

Claims (10)

1. The utility model provides a power station equipment state degradation early warning method which is characterized by comprising the following steps: s1, integrating a charge sensor and a high-precision dielectric sensor in a lubricating oil circulation system of power station equipment; s2, setting an acquisition time interval, dividing a target acquisition time period into acquisition time points according to the preset acquisition time interval, further preliminarily acquiring charge data of the oil products of the power station equipment at the acquisition time points, and identifying an oil product degradation preliminary evaluation coefficient of the lubricating oil corresponding to the power station equipment based on the charge data; S3, based on the preliminary evaluation coefficient of oil degradation of the lubricating oil corresponding to the power station equipment, identifying whether the oil degradation exists in the lubricating oil circulation system corresponding to the power station equipment, if so, executing the step S4, otherwise, jumping out of the method for circulation; And S4, synchronously extracting dielectric data of the oil products of the power station equipment at each acquisition time point based on the distributed high-precision dielectric sensors, deeply evaluating the oil product state of the corresponding lubricating oil circulation system of the power station equipment, and sending corresponding early warning signals.
2. The method of claim 1, wherein the electrical charge data of the power plant oil at each collection time point in step S2 includes a charge density, a charge distribution uniformity, a charge change rate and a charge fluctuation amplitude.
3. The method of claim 1, wherein in step S2, the step of identifying the preliminary evaluation coefficient of oil deterioration of the lubricating oil corresponding to the power plant equipment comprises the following steps: S21, adopting a z-score standardized processing mode to sequentially preprocess the charge density, the charge distribution uniformity, the charge change rate and the charge fluctuation amplitude of the power station equipment oil product at each acquisition time point, and training the SOM model until the SOM model converges; S22, mapping the charge density of the preprocessed power station equipment oil product at each acquisition time point into a SOM model after training is completed, obtaining a weight vector of the neuron in the SOM model mapped by the charge density of the preprocessed power station equipment oil product at each acquisition time point, and marking the weight vector as a weight vector of the charge density mapping neuron at each acquisition time point, namely alpha h , wherein H is the number of each acquisition time point, the value range of H is 1 to H, and H is the total number of the acquisition time points; S23, the charge data of the power station equipment corresponding to each normal time point in the normal operation state are derived from the equipment database, so that the charge density of the power station equipment corresponding to each normal time point in the normal operation state is obtained, the charge density is input into an SOM model in the same way, the weight vector of the charge density mapping neuron of the power station equipment corresponding to each normal time point in the normal operation state is obtained, and then a normal state area is formed; s24, calculating Euclidean distance between the charge density mapping neuron of each acquisition time point and the normal state region Alpha i is the weight vector of the ith neuron in the normal state area, i is the number of each neuron in the normal state area, i=1, 2,..N, N is the total number of neurons in the normal state area, and distance (·) is a distance calculation function; s25, according to an analysis formula Analyzing a change floating coefficient SL1 of the charge density of the power station equipment corresponding to the lubricating oil in a target acquisition time period, wherein D is a set reference charge density value; S26, calculating the change floating coefficient of the charge distribution uniformity, the charge change rate and the charge fluctuation amplitude of the lubricating oil corresponding to the power station equipment in the target acquisition time period according to the calculation mode of the change floating coefficient of the charge density of the lubricating oil corresponding to the power station equipment in the target acquisition time period; S27, finally identifying an oil product degradation preliminary evaluation coefficient psi of the lubricating oil corresponding to the power station equipment, wherein a specific identification formula is as follows In the formula, SL2, SL3 and SL4 respectively represent the charge distribution uniformity, charge change rate and change floating coefficient of the charge fluctuation range of the lubricating oil corresponding to the power station equipment in the target acquisition time period, SL1', SL2', SL3 'and SL4' respectively represent the set charge density, charge distribution uniformity, charge change rate and maximum allowable change floating coefficient of the charge fluctuation range, and omega 1, omega 2, omega 3 and omega 4 respectively represent the distribution weight factors corresponding to the charge density, charge distribution uniformity, charge change rate and charge fluctuation range and reflect the importance of each charge data in the preliminary evaluation of oil degradation.
4. The method according to claim 3, wherein in step S21, training the SOM model until the SOM model converges comprises the steps of: S211, selecting a two-dimensional grid to represent an SOM model, and randomly initializing a weight vector of each neuron in the SOM model; S212, randomly selecting one sample x (t) from the standardized data set as input data to be input into the SOM model, calculating the distances between the sample and all nodes in the SOM model, and taking the node with the smallest distance as the best matching node of the sample x (t): Wherein, the argmin (·) function represents that a parameter value which minimizes a certain function is found, w j (t) is a weight vector of a node j, j is the number of each node, and BMU is the best matching node of a sample x (t); s213, successively updating weight vectors of all nodes in the SOM model, wherein a one-time updating formula is as follows: Wherein w j (t+1) represents the weight vector of the node j after one update, β1 (t) represents the learning rate, β2 bj (t) represents the neighborhood function, exp (·) is the exponential function, γ b 、γ j represents the grid coordinates of the best matching node and the node j, and σ (t) is the neighborhood width; and S214, repeating iterative updating, gradually reducing the learning rate and the neighborhood width until the two-dimensional grid converges, and indicating that the SOM model converges when the two-dimensional grid converges.
5. The method of claim 3, wherein in step S23, forming the normal state neuron area comprises the steps of: s231, extracting weight vectors of all neurons in a normal state area from the SOM model, and summarizing the weight vectors to form a data set; S232, determining a clustering number R through an elbow rule; S233, randomly selecting R weight vectors as initial clustering centers; s234, calculating Euclidean distance between each neuron weight vector alpha i and each clustering center; S235, distributing each neuron weight vector into a cluster center with the minimum Euclidean distance according to the Euclidean distance between each neuron weight vector and each cluster center; S236, carrying out iterative updating on each cluster center, wherein the specific updating step comprises the steps of acquiring the number of internal neurons, simultaneously updating each cluster center, summing the neuron weight vectors of each cluster center, and then taking an average value to recalculate the central point of each cluster; S237, repeatedly carrying out iterative updating on each cluster center until the central point of each cluster center is not changed or is smaller than a set threshold value, and indicating that each cluster center is converged; And S238, after the clustering is completed, identifying a neuron clustering center area in the normal state area, and summarizing the neuron clustering center area to obtain the normal state neuron area.
6. The method according to claim 1, wherein in step S3, the logic for identifying whether the oil product degradation exists in the power station equipment corresponding lubricating oil circulation system is that the oil product degradation preliminary evaluation coefficient of the power station equipment corresponding lubricating oil is compared with a preset oil product degradation threshold coefficient, and if the oil product degradation preliminary evaluation coefficient of the power station equipment corresponding lubricating oil is larger than the preset oil product degradation threshold coefficient, the oil product degradation exists in the power station equipment corresponding lubricating oil circulation system is judged.
7. The method of claim 1, wherein the dielectric data of the power plant oil at each acquisition time point in step S4 includes dielectric constant, dielectric loss tangent, dielectric strength and dielectric response time.
8. The method according to claim 1, wherein in step S4, the deep evaluation of the oil state of the power station equipment corresponding to the lubricating oil circulation system specifically comprises the following steps: S41, extracting dielectric constants, dielectric loss factors, dielectric intensities and dielectric response times of the power station equipment oil products at all acquisition time points, preprocessing the dielectric constants, the dielectric loss factors, the dielectric intensities and the dielectric response times of the power station equipment oil products at all acquisition time points after preprocessing, and marking the preprocessed power station equipment oil products as delta 1 (h), delta 2 (h), delta 3 (h) and delta 4 (h) respectively; S42, dielectric representation of each acquisition time point is represented as a vector X (h) = [ delta 1 (h), delta 2 (h), delta 3 (h), delta 4 (h) ], and then the mahalanobis distance MS (h) between dielectric data and normal dielectric data average value of the oil product of the power station equipment at each acquisition time point is calculated; s43, carrying out statistical analysis according to a time sequence of the acquisition time point, and extracting the degree of the deviation of the whole current oil product from a normal state and the change rate of the oil product degradation trend; S44, finally defining the oil degradation index of the corresponding lubricating oil circulation system of the power station equipment as follows: In the formula, The degree of the deviation of the whole current oil product from the normal state is that theta is the change rate of the oil product degradation trend of the oil product of power station equipment in the target acquisition time period; s45, comparing an oil product degradation index of the power station equipment corresponding to the lubricating oil circulation system with a set oil product degradation standard interval [ Q1, Q2], wherein Q1 is a set oil product degradation allowable lower limit value, and Q2 is a set oil product degradation allowable upper limit value; If it is The oil product state of the power station equipment corresponding to the lubricating oil circulation system is indicated to be in a normal state; If it is The oil product state of the corresponding lubricating oil circulation system of the power station equipment is slightly deteriorated; If it is It indicates that the oil state of the power plant equipment corresponding to the lubricating oil circulation system is severely deteriorated.
9. The method of claim 8, wherein in step S42, calculating the Mahalanobis distance between the dielectric data of the power plant oil at each acquisition time point and the mean value of the normal dielectric data comprises the following steps: S421, dielectric data of the power station equipment at each normal time point in a normal operation state are derived from the equipment database, mean value calculation is carried out on the dielectric data to obtain a dielectric data mean value of the power station equipment in the normal operation state, the dielectric data mean value is similarly formed into a dielectric data vector X ' = [ delta 1', delta 2', delta 3', delta 4' ] of the power station equipment in the normal operation state, and a normal covariance matrix of the power station equipment in the normal operation state is constructed: Wherein δ1' p 、δ2′ p 、δ3′ p 、δ4′ p respectively represents the dielectric constant, the dielectric loss factor, the dielectric strength and the dielectric response time of the power station equipment at the p-th normal time point under the normal operation state, var (-) represents the variance calculation function, cov (-) represents the covariance calculation function, and p is the number of each normal time point; s422, calculating the mahalanobis distance between the dielectric data of the oil product of the power station equipment at each acquisition time point and the average value of the normal dielectric data: In the formula, T is denoted as a transposed symbol.
10. The method according to claim 8, wherein in step S43, the step of extracting the degree of deviation of the current oil product from the normal state and the change rate of the oil product degradation trend comprises the steps of: s431, through a calculation formula Calculating the degree of the deviation of the whole current oil product from the normal state S432, synchronously calculating the change rate of the oil degradation trend of the power station equipment oil product in the target acquisition time period: Wherein Δt is a preset acquisition time interval, and t h is a time value of the h acquisition time point.

Description

Power station equipment state degradation early warning method Technical Field The invention belongs to the technical field of power station equipment state early warning, and particularly relates to a power station equipment state degradation early warning method. Background The oil product monitoring of the power station equipment has important significance and necessity for guaranteeing the safe operation of the equipment and prolonging the service life. Lubricating oil and insulating oil widely used in power station equipment not only play roles in lubrication, cooling and insulation, but also directly influence the operation efficiency and safety of the equipment. As the operating time of the equipment increases, the oil product may be gradually deteriorated by the influence of external environments (such as high temperature, moisture, impurities, etc.), resulting in a decrease in lubrication performance, a decrease in insulation performance, and even possibly causing equipment failure or safety accidents. Through the regular monitoring of the oil product, the degradation trend of the oil product can be found in time, and measures are taken in advance to replace or treat the oil product, so that the serious faults of equipment due to the problem of the oil product are avoided. In addition, the oil product monitoring can also provide important reference for the running state of the equipment, the running condition and potential hidden trouble of the equipment can be indirectly reflected through the change of oil product parameters, the state maintenance and the fine operation and maintenance are facilitated, the operation and maintenance cost is reduced, and the reliability and the economy of the equipment are improved. Therefore, oil product monitoring is one of key links for guaranteeing long-period stable operation of power station equipment. However, the current technology still has some technical defects and shortcomings in the aspect of monitoring the oil products of power station equipment, and the comprehensiveness and the accuracy of monitoring are restricted. First, conventional oil monitoring generally relies on laboratory analysis, requires manual sampling for inspection, has a long period, and cannot realize real-time on-line monitoring, and this hysteresis may cause degradation problems to damage equipment before the detection result comes out. Secondly, the existing oil monitoring technology mainly uses single indexes such as viscosity, acid value or breakdown voltage, but the indexes can only reflect a certain aspect of oil degradation, lack comprehensive evaluation on the comprehensive performance of the oil, and are easy to miss potential problems. Disclosure of Invention The main object of the present invention is to provide a power station equipment state degradation early warning method, aiming at the above mentioned problems. For this purpose, the above object of the present invention is achieved by the following technical solutions: A power station equipment state degradation early warning method comprises the following steps: s1, integrating a charge sensor and a high-precision dielectric sensor in a lubricating oil circulation system of power station equipment; s2, setting an acquisition time interval, dividing a target acquisition time period into acquisition time points according to the preset acquisition time interval, further preliminarily acquiring charge data of the oil products of the power station equipment at the acquisition time points, and identifying an oil product degradation preliminary evaluation coefficient of the lubricating oil corresponding to the power station equipment based on the charge data; S3, based on the preliminary evaluation coefficient of oil degradation of the lubricating oil corresponding to the power station equipment, identifying whether the oil degradation exists in the lubricating oil circulation system corresponding to the power station equipment, if so, executing the step S4, otherwise, jumping out of the method for circulation; And S4, synchronously extracting dielectric data of the oil products of the power station equipment at each acquisition time point based on the distributed high-precision dielectric sensors, deeply evaluating the oil product state of the corresponding lubricating oil circulation system of the power station equipment, and sending corresponding early warning signals. The invention can also adopt or combine the following technical proposal when adopting the technical proposal: in step S2, the charge data of the power station oil product at each acquisition time point comprises charge density, charge distribution uniformity, charge change rate and charge fluctuation amplitude. In step S2, the preliminary evaluation coefficient for identifying the oil quality degradation of the lubricating oil corresponding to the power station equipment specifically comprises the following steps: S21, adopting a z-score standardized pro