Hydroelectric generating set bearing bush temperature health degree calculation method

Abstract

The invention provides a hydroelectric generating set bearing bush temperature health degree calculation method, which comprises the following steps: collecting the bearing bush temperatures of n bearing bushes, setting an allowable bearing bush temperature difference as D, and obtaining the temperature of each bearing bush after treatment; calculating the number of the bearing bushes with different bearing bush temperatures; calculating the occurrence probability of the bearing bush with the temperature of k; the information entropy of the tile temperature difference confusion degree at the t moment is calculated; calculating the maximum value Tmax of the temperature of the bearing bush; training a GBDT model by using historical data of the bearing bush; the bearing bush temperature maximum value Tmaxpre at the current moment is predicted through a GBDT model; calculating the temperature residual Tres of the bearing bush at the moment t; and calculating the overall health degree of the bearing bush temperature by using a two-dimensional Gaussian membership function and integrating two dimensions of the bearing bush temperature difference and the bearing bush temperature change trend. The health state of the bearing bush temperature is reflected more comprehensively from the two dimensions of the bearing bush temperature change trend and the bearing bush temperature difference. According to the method, the overall health degree of the bearing bush temperature is calculated, so that the health state of the bearing bush temperature can be expressed in a more comprehensive and more visual manner.

Description

Technical Field

[0001] The invention relates to the field of hydroelectric bearings, and in particular to a method for calculating the temperature health of a hydroelectric generator set bearing. Background Art

[0002] Bearings are important components of hydropower units, and are divided into guide bearings and thrust bearings. Guide bearings mainly bear the radial mechanical and electromagnetic unbalanced forces of the rotating parts of the unit, while thrust bearings mainly bear the axial water thrust of the turbine generator unit and the weight of the rotating parts. The bearing is the heart of the bearing. Large hydropower units generally adopt a block bearing structure. Abnormal bearing temperature will affect the safe and stable operation of the unit, and in severe cases will cause bearing burnout accidents. Therefore, it is particularly important to monitor the health of the bearing. However, the existing bearing temperature health evaluation dimension is single, and the health status of the bearing temperature is judged only from one dimension of temperature change trend, without considering the impact of the temperature difference of each bearing on the health of the bearing. The temperature difference between different bearings may be caused by many factors, such as uneven load distribution and inconsistent cooling effect. These differences not only affect the life and performance of the bearing itself, but may also pose a potential threat to the operating stability of the entire equipment, resulting in a non-standard evaluation of the bearing temperature health status. At the same time, the bearing temperature cannot be quantified, and the bearing temperature health status is not expressed in a more comprehensive and intuitive way. For this reason, we propose a method for calculating the bearing temperature health status of a hydropower unit to solve the above problems.

[0003] Chinese patent document CN115185313A records a trend tracking and early warning method and device for the bearing temperature of a hydropower unit, including constructing a healthy temperature rise curve of the equipment through methods such as neural networks, and determining the current state of the equipment by comparing the actual temperature rise curve of the equipment with the healthy temperature rise curve (if the actual temperature rise curve is below the healthy temperature rise curve, the equipment is normal; if the actual temperature rise curve crosses or is above the healthy temperature rise curve, the equipment is abnormal). However, this method only judges the healthy state of the bearing temperature from one dimension of the temperature change trend, and the healthy state evaluation of the bearing temperature is not standard, and there are defects in its use, which needs to be improved. Summary of the invention

[0004] The present invention provides a method for calculating the temperature health of the bearing of a hydropower unit, which solves the problem that the evaluation dimension is single and the health status of the bearing temperature is judged only from one dimension of temperature change trend, without considering the influence of the temperature difference of each bearing on the health status of the bearing.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for calculating the temperature health of a bearing of a hydropower unit, comprising the following steps: S1, collect the bearing temperatures of n bearings, set the allowable bearing temperature difference to D, and obtain the temperature of each bearing after processing; S2. Calculate the number of bearings with different bearing temperatures; S3. Calculate the probability of the bearing appearing at a temperature of k; S4, calculate the information entropy of the degree of confusion of the temperature difference of the watt temperature at time t; S5. Calculate the maximum bearing temperature Tmax; S6. Use the historical data of the bearing to train the GBDT model; S7, using the GBDT model to predict the maximum bearing temperature Tmax_pre at the current moment; S8, calculate the bearing temperature residual Tres at time t; S9. Use the two-dimensional Gaussian membership function to comprehensively consider the bearing temperature difference and the bearing temperature change trend to calculate the overall health of the bearing temperature.

[0006] In the preferred solution, in S1, the bearing temperatures T1, T2, ..., Tn of n bearings are collected to obtain the bearing temperature after processing; The specific calculation method of the i-th bearing temperature Ti is: Ti_= , i=1,2,......,n; The temperature of the i-th bearing is Ti, i=1,2,......,n.

[0007] In the preferred solution, in S2, the number of bearings N_i with bearing temperature x_i=k_i is calculated, and N_i≤n, where k_i∈(T1_,T2_,......,Tn_), i=1,2,......,m, m is the number of bearings with different temperatures and m≤n, .

[0008] In the preferred solution, the probability P in S3 is calculated as follows: ; The probability of a bearing shell with temperature k appearing.

[0009] In the preferred solution, the information entropy H(t) in S4 is calculated as follows: ; in, is the information entropy reflecting the degree of disorder of temperature difference at time t, P( ) is the probability of the bearing having a bearing temperature of k at time t.

[0010] In the preferred solution, in S6, the GBDT model is trained using historical data of maximum active power, water head, rotation speed, oil level, oil temperature, swing, ambient temperature and bearing temperature.

[0011] In the preferred solution, the calculation method of the bearing temperature residual at time t in S8 is: Tres(t)=Tmax_pre(t)-Tmax(t); Tmax_pre(t) is the maximum bearing temperature predicted by the GBDT model at time t, Tmax(t) is the maximum bearing temperature at time t, and Tres(t) is the residual bearing temperature at time t.

[0012] In the preferred solution, the calculation method of the two-dimensional Gaussian function in S9 is as follows: A1: Calculation method of two-dimensional Gaussian function: ; in, , They are the parameters of the reaction function image fatness and thinness, , Represents the center point of the function graph; A2: Substitute the information entropy and the bearing temperature residual into the two-dimensional Gaussian function to calculate the bearing temperature health. The calculation formula for the bearing temperature health is as follows: ; =0, is the allowed residual, , are the parameters that need to be adjusted.

[0013] The beneficial effects of the present invention are as follows: the present invention more comprehensively reflects the health status of the bearing temperature from two dimensions: the bearing temperature change trend and the bearing temperature difference. The present invention uses the information entropy principle to reflect the overall disorder degree of the bearing temperature, so that the change of the bearing temperature difference can be quantitatively characterized. The present invention uses the GBDT algorithm to predict the maximum value of the bearing temperature, and captures the source of the bearing temperature degradation through the change trend of the maximum bearing temperature. The present invention uses a two-dimensional Gaussian membership function to combine the information entropy of the bearing temperature difference and the residual obtained by the GBDT algorithm to calculate the overall health of the bearing temperature, so that the health status of the bearing temperature can be expressed in a more comprehensive and intuitive way, which has great promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below in conjunction with the accompanying drawings and embodiments; Figure 1 It is a flow chart of the temperature health calculation method of the present invention; DETAILED DESCRIPTION Embodiment 1: like Figure 1 A method for calculating the temperature health of a bearing of a hydropower unit comprises the following steps: S1, collect the bearing temperatures of n bearings, set the allowable bearing temperature difference to D, and obtain the temperature of each bearing after processing; S2. Calculate the number of bearings with different bearing temperatures; S3. Calculate the probability of the bearing appearing at a temperature of k; S4, calculate the information entropy of the degree of confusion of the temperature difference of the watt temperature at time t; S5. Calculate the maximum bearing temperature Tmax; S6. Use the historical data of the bearing to train the GBDT model; S7, using the GBDT model to predict the maximum bearing temperature Tmax_pre at the current moment; S8, calculate the bearing temperature residual Tres at time t; S9. Using a two-dimensional Gaussian membership function, the overall health of the bearing temperature is calculated by integrating the two dimensions of bearing temperature difference and bearing temperature change trend. According to this method, the smaller the bearing temperature difference is in principle, the better. By controlling the temperature difference within a reasonable range, the safe temperature operation of the unit can be met. This method first rounds the bearing temperature to the nearest integer, and then calculates the probability of occurrence of different bearing temperatures. The larger the probability obtained each time, the more concentrated the bearing temperature is at that temperature, and vice versa. The present invention introduces information entropy to characterize the dispersion and concentration of the bearing temperature.

[0015] The present invention more comprehensively reflects the health status of the bearing temperature from two dimensions: the bearing temperature change trend and the bearing temperature difference. The present invention uses the information entropy principle to reflect the overall disorder degree of the bearing temperature, so that the change of the bearing temperature difference can be quantitatively characterized. The present invention uses the GBDT algorithm to predict the maximum value of the bearing temperature, and captures the source of the bearing temperature degradation through the change trend of the maximum bearing temperature. The present invention uses a two-dimensional Gaussian membership function to combine the information entropy of the bearing temperature difference and the residual obtained by the GBDT algorithm to calculate the overall health of the bearing temperature, so that the health status of the bearing temperature can be expressed in a more comprehensive and intuitive way.

[0016] In the preferred solution, in S1, the bearing temperatures T1, T2, ..., Tn of n bearings are collected to obtain the bearing temperature after processing; The specific calculation method of the i-th bearing temperature Ti is: Ti_= , i=1,2,......,n; The temperature of the i-th bearing is Ti, i=1,2,......,n.

[0017] In the preferred solution, in S2, the number of bearings N_i with bearing temperature x_i=k_i is calculated, and N_i≤n, where k_i∈(T1_,T2_,......,Tn_), i=1,2,......,m, m is the number of bearings with different temperatures and m≤n, .

[0018] In the preferred solution, the probability P in S3 is calculated as follows: ; The probability of a bearing shell with temperature k appearing.

[0019] In the preferred solution, the information entropy H(t) in S4 is calculated as follows: ; in, is the information entropy reflecting the degree of disorder of temperature difference at time t, P( ) is the probability of the bearing with a bearing temperature of k at time t. In this way, entropy is a measure and scale of the degree of chaos in a system. When the ordered state of the system is consistent, the entropy value is smaller where the data is more concentrated, and the entropy value is larger where the data is more dispersed. Description from the orderliness of information: when the amount of data is consistent, the more ordered the system is, the lower the entropy value is; the more chaotic or dispersed the system is, the higher the entropy value is. Information entropy is a basic concept in information theory, which describes the uncertainty of possible events in the information source.

[0020] In the preferred solution, in S6, the GBDT model is trained using historical data of active power, water head, speed, oil level, oil temperature, swing, ambient temperature and maximum bearing temperature. In this way, the bearing temperature degradation or over-limit often starts from the highest bearing temperature. Therefore, this method uses the GBDT algorithm to predict the maximum bearing temperature through quantities related to bearing temperature (such as active power, water head, speed, oil level, oil temperature, swing, ambient temperature, etc.).

[0021] GBDT, also known as gradient boosted decision tree, is an iterative decision tree algorithm that consists of multiple decision trees, and the conclusions of all trees are accumulated to get the final answer. The GBDT algorithm can be regarded as an additive model composed of M trees.

[0022] In the preferred solution, historical health samples of active power, water head, speed, oil level, oil temperature, swing, and ambient temperature are ; in, , ,......, They are historical healthy samples of active power, water head, speed, oil level, oil temperature, swing and ambient temperature.

[0023] Historical sample data on health Perform value compensation and frequency reduction processing, use active power as the judgment condition, select data when the unit is in stable operation, and remove empty values ​​and abnormal points in the data.

[0024] Normalize the input parameters. The normalization method is: ; The final healthy sample data is obtained through the above processing , , , , , ].

[0025] Based on the GBDT algorithm model, the above healthy samples are used to train the bearing temperature maximum prediction model, and the The predicted value of the maximum bearing temperature at time Tmax_pre(t).

[0026] The calculation method of the bearing temperature residual at time t in S8 is: Tres(t)=Tmax_pre(t)-Tmax(t); Tmax_pre(t) is the maximum bearing temperature predicted by the GBDT model at time t, Tmax(t) is the maximum bearing temperature at time t, and Tres(t) is the residual bearing temperature at time t.

[0027] In the preferred solution, the calculation method of the two-dimensional Gaussian function in S9 is as follows: A1: Calculation method of two-dimensional Gaussian function: ; A2: Substitute the information entropy and the bearing temperature residual into the two-dimensional Gaussian function to calculate the bearing temperature health. The calculation formula for the bearing temperature health is as follows: ; =0, is the allowed residual, , are the parameters that need to be adjusted.

[0028] Embodiment 2: Further explanation in combination with Example 1: The bearing temperature difference treatment method is as follows: The calculation formula of information entropy H(x) is as follows: (1) Bearing temperature data preprocessing: Assume that the bearing temperatures of n bearings are T1, T2, ..., Tn, and the allowable bearing temperature difference is D. Then the temperature of the i-th bearing after processing is Ti.

[0029] Ti_= , i=1,2,......,n; (2) Calculate the number of bearings N_i with bearing temperature x_i=k_i, where N_i≤n, where k_i∈(T1_,T2_,......,Tn_), i=1,2,......,m, m is the number of bearings with different temperatures and m≤n, .

[0030] (3) Calculate the probability of the bearing appearing at a temperature of k ; ; (4) Calculate the information entropy H(t) at time t:

[0031] Embodiment 3: Further explanation in combination with Examples 1 and 2: The bearing temperature difference treatment method is as follows: Bearing temperature prediction method: GBDT, also known as gradient boosted decision tree, is an iterative decision tree algorithm that consists of multiple decision trees, and the conclusions of all trees are accumulated to get the final answer. The GBDT algorithm can be regarded as an addition model composed of M trees, and its corresponding formula is as follows: ; Among them, x is the input sample, w is the model parameter, h is the classification regression tree, and α is the weight of each tree.

[0032] The GBDT algorithm uses the negative gradient of the loss function in the current model as an approximation of the residual of the boosted tree model: ; in, is the loss function and r is the residual.

[0033] The bearing temperature prediction method is as follows: (1) Calculate the maximum bearing temperature Tmax; (2) GBDT model training. The GBDT model is trained using historical data such as active power, water head, speed, oil level, oil temperature, swing, ambient temperature, and maximum bearing temperature. (3) GBDT model prediction. Use the GBDT model to predict the maximum bearing temperature Tmax_pre at the current moment; (4) Calculate the bearing temperature residual Tres at time t: Tres(t)=Tmax_pre(t)-Tmax(t).

[0034] Embodiment 4: Further explanation in combination with Examples 1 to 3: The specific steps for calculating the temperature health of the bearing bush are as follows: The two-dimensional Gaussian membership function is used to integrate the two dimensions of bearing temperature difference and bearing temperature change trend to calculate the overall health of the bearing temperature.

[0035] The calculation method of the two-dimensional Gaussian function is as follows: ; in, , They are the parameters of the reaction function image fatness and thinness, , Represents the center point of the function graph.

[0036] The smaller the information entropy reflecting the temperature difference of the tile temperature, the better the consistency of the tile temperature; the smaller the absolute value of the residual reflecting the trend of the tile temperature change, the better the trend of the tile temperature change. In fact, it is impossible to completely eliminate the residual, and it is only necessary to control the residual within the allowable range.

[0037] The bearing temperature health at this moment is calculated as follows: ; in is the information entropy reflecting the degree of disorder of the temperature difference at time t, is the bearing temperature residual at time t, =0, is the allowed residual, , are the parameters that need to be adjusted.

[0038] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A method for calculating the temperature health of a bearing of a hydropower unit, characterized by: The following steps are involved: S1, collect the bearing temperatures of n bearings, set the allowable bearing temperature difference to D, and obtain the temperature of each bearing after processing; S2. Calculate the number of bearings with different bearing temperatures; S3. Calculate the probability of the bearing appearing at a temperature of k; S4, calculate the information entropy of the degree of confusion of the temperature difference of the watt temperature at time t; S5. Calculate the maximum bearing temperature Tmax; S6. Use the historical data of the bearing to train the GBDT model; S7, using the GBDT model to predict the maximum bearing temperature Tmax_pre at the current moment; S8, calculate the bearing temperature residual Tres at time t; S9. Use the two-dimensional Gaussian membership function to comprehensively consider the bearing temperature difference and the bearing temperature change trend to calculate the overall health of the bearing temperature.

2. According to claim 1, a method for calculating the temperature health of a hydropower unit bearing, characterized in that: In S1, the bearing temperatures of n bearings T1, T2, ..., Tn are collected to obtain the bearing temperature after processing; The specific calculation method of the i-th bearing temperature Ti is: You_= ,i=1,2,......,n; The temperature of the i-th bearing is Ti, i=1,2,......,n.

3. According to claim 1, a method for calculating the temperature health of a bearing of a hydropower unit is characterized in that: in S2, the number of bearings N_i with bearing temperature x_i=k_i is calculated, and N_i≤n, where k_i∈(T1_,T2_,......,Tn_), i=1,2,......,m, m is the number of bearings with different temperatures and m≤n, .

4. According to the method for calculating the temperature health of the bearing of a hydropower unit according to claim 1, it is characterized by: S3 The calculation method of the probability P is: ； The probability of a bearing shell with temperature k appearing.

5. According to the method for calculating the temperature health of the bearing of a hydropower unit according to claim 1, the characteristics are: S4 The information entropy H(t) is calculated as follows: ； in, is the information entropy reflecting the degree of disorder of temperature difference at time t, P( ) is the probability of the bearing having a bearing temperature of k at time t.

6. A method for calculating the temperature health of a bearing of a hydropower unit according to claim 1, characterized in that: in S6, the GBDT model is trained using historical data of active power, water head, speed, oil level, oil temperature, swing, ambient temperature and maximum bearing temperature.

7. According to claim 1, a method for calculating the temperature health of a bearing of a hydropower unit is characterized by: The calculation method of the bearing temperature residual at time t in S8 is: Tres(t)=Tmax_pre(t)-Tmax(t); Tmax_pre(t) is the maximum bearing temperature predicted by the GBDT model at time t, Tmax(t) is the maximum bearing temperature at time t, and Tres(t) is the residual bearing temperature at time t.

8. According to claim 1, a method for calculating the temperature health of a bearing of a hydropower unit is characterized by: The calculation method of the two-dimensional Gaussian function in S9 is as follows: A1: Calculation method of two-dimensional Gaussian function: ； in, , They are the parameters of the reaction function image fatness and thinness, , Represents the center point of the function graph; A2: Substitute the information entropy and the bearing temperature residual into the two-dimensional Gaussian function to calculate the bearing temperature health. The calculation formula for the bearing temperature health is as follows: ； =0, is the allowed residual, ax and ay are the parameters that need to be adjusted.

Images