CN109443419B - Machine learning-based rectifier online monitoring method - Google Patents
Machine learning-based rectifier online monitoring method Download PDFInfo
- Publication number
- CN109443419B CN109443419B CN201811011319.9A CN201811011319A CN109443419B CN 109443419 B CN109443419 B CN 109443419B CN 201811011319 A CN201811011319 A CN 201811011319A CN 109443419 B CN109443419 B CN 109443419B
- Authority
- CN
- China
- Prior art keywords
- current
- value
- rectifier
- time
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Tests Of Circuit Breakers, Generators, And Electric Motors (AREA)
Abstract
A machine learning-based rectifier online monitoring method comprises the following steps: modeling by using historical data of the rectifier, and constructing a temperature model of the rectifier in a normal operation state, wherein the temperature model comprises links of data processing and machine learning; rectifier state monitoring and prediction: processing new data by using data processing, constructing the same characteristic variables, transmitting the processed data to a rectifier temperature model, and calculating a predicted value of the rectifier temperature by using the temperature model; rectifier state/early warning level calculation: comparing the predicted value with the true value of the rectifier temperature in real time, judging the running state of the rectifier by adopting interval judgment and dynamic monitoring, and giving out early warning information and state information of the rectifier in real time, wherein the early warning information is divided into three intervals: the early warning level of the first interval is 0, which indicates that the equipment normally operates; the early warning level of the second interval is 1-11, which indicates that the equipment state needs to be 'noticed'; the second interval warning level is 12, indicating that the device status is "abnormal".
Description
Technical Field
The invention relates to a machine learning-based online rectifier monitoring method.
Background
The Rectifier (Rectifier) is a rectifying device which can convert Alternating Current (AC) into Direct Current (DC), and the equipment mainly has two functions, namely, converting the Alternating Current (AC) into the Direct Current (DC), and supplying the filtered Alternating Current (AC) to a load; and secondly, supplying power to the energy storage system. As a part of the power supply system, the running state of the rectifier is directly related to the stability of the power supply system, and the safety of the power supply system during running can be further improved by monitoring, judging and early warning the running state of the rectifier equipment.
So far, the monitoring mode of the rectifier running state at home and abroad mainly collects the current, the environment temperature, the environment humidity and the equipment temperature data when the equipment runs and displays the data through numerical values. However, the conventional monitoring method cannot describe the mutual influence relationship between the variables, and on the other hand, an index threshold (e.g., an equipment temperature threshold) must be manually set when the rectifier state is evaluated, and when the index is within the threshold, the overall state and the variation trend of the rectifier cannot be accurately and effectively evaluated.
Disclosure of Invention
The invention aims to solve the technical problem of providing a machine learning-based rectifier online monitoring method, which takes a rectifier real temperature value as a state judgment index, constructs a temperature model between the rectifier current, power, environment temperature and humidity and the rectifier real temperature, uses the temperature model to calculate a predicted value of the rectifier temperature in real time when monitoring the rectifier state, compares an equipment predicted temperature value with the real temperature value, and discriminates the abnormal operation trend of the rectifier in real time by judging the result, so as to monitor the operation state of the rectifier.
The technical scheme adopted by the invention is as follows:
a machine learning-based rectifier online monitoring method is characterized by comprising the following steps:
s1, data acquisition:
collecting data of current, equipment temperature, environment temperature and environment humidity when the rectifier runs by using a current sensor, a temperature sensor and a humidity sensor;
s2, data processing: the current data is subjected to dynamic smoothing treatment, and the method comprises a moving smoothing method, an asymmetric local weighted regression scatter point smoothing method and a moving window simulation or polynomial smoothing method;
s2-1, performing dynamic smoothing processing on the current data by adopting a moving smoothing method, namely a formula (1); because the current change has large fluctuation, the result fluctuation is large when the current is directly used for prediction; therefore, the current data needs to be smoothed before modeling and temperature prediction, so that the predicted temperature is ensured not to have large fluctuation;
in the invention, a moving smoothing method is adopted to carry out moving smoothing treatment on current data:
wherein the content of the first and second substances,
for the current value smoothed at the present moment, ItAs the actual current value at the current time, (1,2, …, N) is how many time points, such as I, are smoothed backwards at the current timet-1Is the current value at time t-1, It-2Is the current value at time t-2, It-NThe current value at the time t-N;
s2-2, constructing a new derivative variable by using the current data:
constructing derivative variables of the current data, wherein the derivative variable types comprise a current square value, a current cubic value, a smooth current square value, a smooth current cubic value, a current square accumulated value in a period of time (formula (2)), a current cubic accumulated value in a period of time (formula (3)), a current maximum (formula (4)), a minimum value (formula (5)) and a standard deviation (formula (6));
It_max＝max(It-1,It-2,…,It-N)………………………(4)；
It_min＝min(It-1,It-2,…,It-N)………………………(5)；
It_std＝std(It-1,It-2,…,It-N)………………………(6)；
in the formula, the first and second sets of data are represented,
is the square of the current at time t,
the square of the current at time t-1,
the square of the current at time t-2,
the square value of the current at the time t-N;
current at time tThe value of the cube is calculated according to the formula,
is the current cubic value at time t-1,
is the current cubic value at time t-2,
the current cubic value at the time t-N;
is [ t-N, t-1 ]]The integrated value of the square of the current over the time period,
is [ t-N, t-1 ]]Cumulative value of current cubed over time period, It_maxIs [ t-N, t-1 ]]Maximum value of current in time period, It_minIs [ t-N, t-1 ]]Minimum value of current in time period, It_stdIs [ t-N, t-1 ]]Standard deviation of current values over a time period; i ist-1Is the current value at time t, It-2Is the current value at time t-2, It-NThe current value at the time t-N;
s2-3, data normalization:
scaling the data to make the data fall into a cell, eliminating unit limitation of the data, and facilitating comparison and weighting of indexes of different units or orders of magnitude;
in the present invention, data normalization is performed for each variable using the z-score method (normalization method) (equation 9):
σX＝std(x1,x2,…,xN)………………………(8)；
in addition to the z-score method, it may be: min-max normalization, log function conversion, and atan function conversion methods;
s3, machine learning: modeling by using a ridge regression method, a Lasso regression method, a random forest, a decision tree, a gradient boosting decision tree, a neural network or an RNN algorithm according to historical operation data of a rectifier;
the method adopts rectifier historical operation data and a ridge regression method to carry out modeling, and establishes a corresponding relation among current, environment temperature, environment humidity and equipment temperature when the rectifier operates, namely a rectifier temperature model;
for the linear regression problem, the objective function of the least squares method is (equation 10):
in the formula:
xij——xi＝(xi1,…,xip)Tthe independent variable value of the ith observation sample is the independent variable value of the ith observation sample, and the sample has p characteristic variables;
yi-dependent value of the ith sample (i.e. the required actual temperature value of the device);
α -intercept term;
βj-coefficients of the jth characteristic variable;
beta is a vector formed by characteristic variable coefficients;
let θ be (α, β), the objective function becomes to minimize | | X θ -y | | survival2(ii) a Derivation of the parameters of the objective function yields the solution formula for the objective function (formula 11):
θ＝(XTX)-1XTy………………………(11)；
in the formula, theta is a parameter vector to be solved, X is a sample characteristic matrix, and y is a dependent variable value vector; when X is not column full rank, or multiple collinearity exists between columns, XTLine of XColumn wise close to 0 (i.e. X)TX is close to singular), then (X) is calculatedTX)-1The time error is very large, so that the traditional least square method is lack of stability and reliability;
therefore, by abandoning the unbiased property of the least square method, the regression coefficient obtained at the cost of losing part of information and reducing precision is more consistent with the actual and reliable regression method:
adding a regularization term to the target function to minimize | | X θ -y | | calculation of the non-calculation2+||λI||2；
After derivation is performed on the regularized target function, a target function solving formula (formula 12) can be obtained, and the method is called a ridge regression method;
θ＝(XTX+λI)-1XTy………………………(12)；
in the formula, λ is a ridge parameter, I is an identity matrix, and λ I is an increasing regular term;
s4, model prediction:
aiming at new data generated by the operation of the rectifier, the data is transmitted to a rectifier temperature model after being processed, the temperature of the rectifier under the current, the ambient temperature and the ambient humidity is predicted, and whether the temperature of the rectifier changes according to a historical normal rule or not is discriminated in real time by comparing the real temperature and the real temperature of the rectifier;
s5, calculating the state and early warning level of the rectifier
Because the current fluctuation is large, the prediction result of the model at partial data points is poor, and in order to avoid program misinformation, the mode of interval judgment and dynamic monitoring is used for distinguishing on the basis of model prediction;
the steps of the interval determination and dynamic monitoring process are specifically as follows (see fig. 3 for details):
4) judging whether the absolute value of the residual error between the predicted temperature and the real temperature of the current equipment is greater than a set threshold value, and when the absolute value of the residual error is greater than the threshold value, automatically starting a monitoring program to monitor the state of the equipment;
5) counting the absolute values of the residual errors of the subsequent N data points, and when the absolute values of the residual errors of the N data points are larger than a threshold value, automatically adding 1 to the state information Tp of the equipment;
6) repeating the step (2), and when the absolute value of the residual error of n data points appearing in all the continuous p intervals is larger than or equal to the threshold (namely Tp is p), automatically adding 1 to the early warning information Tw; and (4) if the absolute value of the residual error in a certain interval does not meet the condition, resetting the Tp and Tw values, returning to the step (1), and restarting to monitor the model prediction result.
Has the advantages that: the technical scheme fundamentally solves the problem that the current simple monitoring mode that the state of the rectifier cannot be effectively evaluated when the early warning state value is manually set and the index value is in the threshold range is solved. The temperature model among the current, the environment temperature and humidity and the rectifier temperature during the normal operation of the rectifier is established by adopting historical operation data, so that the abnormal change trend of the equipment in the subsequent operation process can be discriminated. Meanwhile, by using a dynamic monitoring program and an interval discrimination method, the technical scheme has stronger anti-interference performance and avoids program misinformation.
Drawings
FIG. 1 is a flow chart of the invention;
FIG. 2 is a diagram of real-time monitoring and prediction of rectifier temperature;
FIG. 3 is a flow chart of rectifier state/warning level calculation;
fig. 4 is a diagram of the rectifier status/warning level display.
Detailed Description
To explain technical contents, structural features, and objects and effects of the technical solutions in detail, the following detailed description is given with reference to specific examples and accompanying drawings.
The embodiment of the machine learning-based rectifier online monitoring method comprises the following steps of:
s1, data acquisition:
collecting data of current, equipment temperature, environment temperature and environment humidity when the rectifier runs by using a current sensor, a temperature sensor and a humidity sensor;
s2, data processing:
s2-1, performing dynamic smoothing treatment on the current (formula 1); because the current change has large fluctuation, the result fluctuation is large when the current is directly used for prediction; therefore, the current data needs to be smoothed before modeling and temperature prediction, so that the predicted temperature is ensured not to have large fluctuation;
in the invention, a moving smoothing method is adopted to carry out moving smoothing treatment on current data:
wherein the content of the first and second substances,
for the current value smoothed at the present moment, ItAs the actual current value at the current time, (1,2, …, N) is how many time points, such as I, are smoothed backwards at the current timet-1Is the current value at time t-1, It-2Is the current value at time t-2, It-NThe current value at the time t-N;
in addition, an asymmetric local weighted regression scatter smoothing method (LOWESS), a moving window fitting polynomial smoothing method can also be adopted;
s2-2, constructing a new derivative variable by using the current data:
constructing derivative variables of the current data, wherein the derivative variable types comprise a current square value, a current cubic value, a smooth current square value, a smooth current cubic value, a current square in a period of time (formula (2)), a cubic cumulative value (formula (3)), a current maximum value (formula (4)), a minimum value (formula (5)) and a standard deviation (formula (6));
It_max＝max(It-1,It-2,…,It-N)………………………(4)；
It_min＝min(It-1,It-2,…,It-N)………………………(5)；
It_std＝std(It-1,It-2,…,It-N)………………………(6)；
in the formula, the first and second sets of data are represented,
is the square of the current at time t,
the square of the current at time t-1,
the square of the current at time t-2,
the square value of the current at the time t-N;
is the current cubic value at time t,
is the current cubic value at time t-1,
is the current cubic value at time t-2,
the current cubic value at the time t-N;
is [ t-N, t-1 ]]The integrated value of the square of the current over the time period,
is [ t-N, t-1 ]]Cumulative value of current cubed over time period, It_maxIs [ t-N, t-1 ]]Maximum value of current in time period, It_minIs [ t-N, t-1 ]]Minimum value of current in time period, It_stdIs [ t-N, t-1 ]]Standard deviation of current values over a time period; i ist-1Is the current value at time t, It-2Is the current value at time t-2, It-NThe current value at the time t-N;
s2-3, data normalization:
scaling the data to make the data fall into a small specific interval, eliminating unit limitation of the data, and facilitating comparison and weighting of indexes of different units or magnitude levels;
in the present invention, data normalization is performed for each variable using the z-score method (normalization method) (equation 9):
σX＝std(x1,x2,…,xN)………………………(8)；
s3, machine learning:
modeling by using historical rectifier operation data and a ridge regression method, and constructing a corresponding relation among current, environment temperature, environment humidity and equipment temperature when the rectifier operates, namely a rectifier temperature model;
for the linear regression problem, the objective function of the least squares method is (equation 10):
in the formula:
xij——xi＝(xi1,…,xip)Tthe independent variable value of the ith observation sample is the independent variable value of the ith observation sample, and the sample has p characteristic variables;
yi-dependent value of the ith sample (i.e. the required actual temperature value of the device);
α -intercept term;
βj-coefficients of the jth characteristic variable;
beta is a vector formed by characteristic variable coefficients;
let θ be (α, β), the objective function becomes to minimize | | X θ -y | | survival2(ii) a Derivation of the parameters of the objective function yields the solution formula for the objective function (formula 11):
θ＝(XTX)-1XTy………………………(11)；
in the formula, theta is a parameter vector to be solved, X is a sample characteristic matrix, and y is a dependent variable value vector. When X is not column full rank, or multiple collinearity exists between columns, XTDeterminant of X is close to 0 (i.e. X)TX is close to singular), then (X) is calculatedTX)-1The time error is very large, so that the traditional least square method is lack of stability and reliability;
therefore, by abandoning the unbiased property of the least square method, the regression coefficient obtained at the cost of losing part of information and reducing precision is more consistent with the actual and reliable regression method:
adding a regularization term to the target function to minimize | | X θ -y | | calculation of the non-calculation2+||λI||2；
After derivation is performed on the regularized target function, a target function solving formula (formula 12) can be obtained, and the method is called a ridge regression method;
θ＝(XTX+λI)-1XTy………………………(12)；
in the formula, λ is the ridge parameter, I is the identity matrix, and λ I is the increasing regularization term.
S4, model prediction
Aiming at new data generated by the operation of the rectifier, the data is transmitted to a rectifier temperature model after being processed, the temperature of the rectifier under the current, the ambient temperature and the ambient humidity is predicted, and whether the temperature of the rectifier changes according to a historical normal rule or not is discriminated in real time by comparing the real temperature and the real temperature of the rectifier;
s5, calculating the state and early warning level of the rectifier
Because the current fluctuation is large, the prediction result of the model at partial data points is poor, and in order to avoid program misinformation, the mode of interval judgment and dynamic monitoring is used for distinguishing on the basis of model prediction;
the steps of the interval determination and dynamic monitoring process are specifically as follows (see fig. 3 for details):
7) judging whether the absolute value of the residual error between the predicted temperature and the real temperature of the current equipment is greater than a set threshold value, and when the absolute value of the residual error is greater than the threshold value, automatically starting a monitoring program to monitor the state of the equipment;
8) counting the absolute values of the residual errors of the subsequent N data points, and when the absolute values of the residual errors of the N data points are larger than a threshold value, automatically adding 1 to the state information Tp of the equipment;
9) repeating the step (2), and when the absolute value of the residual error of n data points appearing in all the continuous p intervals is larger than or equal to the threshold (namely Tp is p), automatically adding 1 to the early warning information Tw; and (4) if the absolute value of the residual error in a certain interval does not meet the condition, resetting the Tp and Tw values, returning to the step (1), and restarting to monitor the model prediction result.
The specific flow is shown in fig. 1 and is divided into three steps.
1) Step 1 models rectifier historical data. The method comprises the steps of modeling by using historical data of the rectifier, constructing a temperature model of the rectifier in a normal operation state, processing original data of the rectifier and constructing a characteristic project by using a data processing technology, and constructing the temperature model of the rectifier by using a ridge regression method in a machine learning technology link.
2) Step 2 is rectifier condition monitoring and prediction (see fig. 2 for details). And collecting the state data of the rectifier, and on one hand, carrying out state monitoring. On the other hand, the temperature value of the rectifier in the normal state is predicted by combining the rectifier temperature model constructed in the step 1 and adopting the technical links of data processing and model prediction. Processing the new data by using a data processing technology, constructing the same characteristic variables, transmitting the processed data to a rectifier temperature model, and calculating a predicted value of the rectifier temperature by using the temperature model.
3) Step 3 is rectifier status/warning level calculation (see figure 3 for details). The step uses the technology of a rectifier state and early warning grade calculation link. Comparing the predicted value with the actual value of the rectifier temperature in real time, judging the operation state of the rectifier by adopting interval judgment and dynamic monitoring, and giving out early warning information and state information of the rectifier in real time (see figure 4 in detail). The early warning information is divided into three intervals. The early warning level of the first interval is 0 (green), which indicates that the equipment normally operates. The second interval warning level is 1-11 (yellow), indicating that the equipment status needs "attention". The second interval warning level is 12 (red), indicating that the device status is "abnormal".
Claims (1)
1. A machine learning-based rectifier online monitoring method is characterized by comprising the following steps:
s1, data acquisition:
collecting data of current, equipment temperature, environment temperature and environment humidity when the rectifier runs by using a current sensor, a temperature sensor and a humidity sensor;
s2, data processing: the current data is subjected to dynamic smoothing treatment, and the method comprises a moving smoothing method, an asymmetric local weighted regression scatter point smoothing method and a moving window simulation or polynomial smoothing method;
s3, machine learning: modeling by using a ridge regression method, a Lasso regression method, a random forest, a decision tree, a gradient boosting decision tree, a neural network or an RNN algorithm according to historical operation data of a rectifier;
s4, model prediction:
aiming at new data generated by the operation of the rectifier, the data is transmitted to a rectifier temperature model after being processed, the temperature of the rectifier under the current, the ambient temperature and the ambient humidity is predicted, and whether the temperature of the rectifier changes according to a historical normal rule or not is discriminated in real time by comparing the real temperature and the real temperature of the rectifier;
in the data processing of S2, the dynamic smoothing of the current data by using the moving smoothing method includes the following substeps:
s2-1, formula (1):
wherein the content of the first and second substances,
for the current value smoothed at the present moment, ItAs the actual current value at the current time, (1,2, …, N) is how many time points, such as I, are smoothed backwards at the current timet-1Is the current value at time t-1, It-2Is the current value at time t-2, It-NThe current value at the time t-N;
s2-2, constructing a new derivative variable by using the current data:
constructing derivative variables of the current data, wherein the derivative variables comprise a current square value, a current cubic value, a smooth current square value, a smooth current cubic value, a current square accumulated value in a period of time of formula (2), a current cubic accumulated value in a period of time of formula (3), a current maximum value of formula (4), a current minimum value of formula (5) and a standard deviation of formula (6);
It_max＝max(It-1,It-2,...,It-N)...........................(4)；
It_min＝min(It-1,It-2,...,It-N)...........................(5)；
It_std＝std(It-1,It-2,...,It-N)...........................(6)；
in the formula, the first and second sets of data are represented,
is the square of the current at time t,
the square of the current at time t-1,
the square of the current at time t-2,
the square value of the current at the time t-N;
is the current cubic value at time t,
is the current cubic value at time t-1,
is the current cubic value at time t-2,
the current cubic value at the time t-N;
is [ t-N, t-1 ]]The integrated value of the square of the current over the time period,
is [ t-N, t-1 ]]The cumulative value of the current cube over the time period,It_maxis [ t-N, t-1 ]]Maximum value of current in time period, It_minIs [ t-N, t-1 ]]Minimum value of current in time period, It_stdIs [ t-N, t-1 ]]Standard deviation of current values over a time period; i ist-1Is the current value at time t, It-2Is the current value at time t-2, It-NThe current value at the time t-N;
s2-3, data normalization:
data normalization was performed for each variable using the z-score method of equation 9:
σX＝std(x1,x2，...,xN).........(8)；
in the S3 machine learning, modeling is performed by adopting rectifier historical operation data and a ridge regression method, and the corresponding relation among the current, the environment temperature, the environment humidity and the equipment temperature when the rectifier operates is established, namely a rectifier temperature model is specifically defined as follows;
for the linear regression problem, the objective function of the least squares method is formula 10:
in the formula:
xij——xi＝(xi1,...,xip)Tthe independent variable value of the ith observation sample is the independent variable value of the ith observation sample, and the sample has p characteristic variables;
yi-the dependent value of the ith sample;
α -intercept term;
βj-coefficients of the jth characteristic variable;
beta is a vector formed by characteristic variable coefficients;
let θ be (α, β), the objective function becomes to minimize | | X θ -y | | survival2(ii) a Derivation is performed on the parameters of the objective function to obtain a solution formula 11 of the objective function:
θ＝(XTX)-1XTy...........................(11)；
in the formula, theta is a parameter vector to be solved, X is a sample characteristic matrix, and y is a dependent variable value vector;
when X is not the full rank of the column or multiple collinearity exists between the columns, a regularization term is added to the target function, so that the target function becomes the minimum | | | X theta-y | |2+||λI||2；
After derivation is performed on the regularized target function, a target function solving formula 12 can be obtained, and the method is called a ridge regression method;
θ＝(XTX+λI)-1XTy...........................(12)；
in the formula, λ is a ridge parameter, I is an identity matrix, and λ I is an increasing regular term;
step S5, rectifier state and early warning level calculation:
on the basis of model prediction, a mode of interval judgment and dynamic monitoring is used for judging, and the steps of the interval judgment and dynamic monitoring process are as follows:
1) judging whether the absolute value of the residual error between the predicted temperature and the real temperature of the current equipment is greater than a set threshold value, and when the absolute value of the residual error is greater than the threshold value, automatically starting a monitoring program to monitor the state of the equipment;
2) counting the absolute values of the residual errors of the subsequent N data points, and when the absolute values of the residual errors of the N data points are larger than a threshold value, automatically adding 1 to the state information Tp of the equipment;
3) repeating the step 2), and when the absolute value of the residual error of n data points in all the continuous p intervals is greater than or equal to the threshold value, namely Tp is equal to p, automatically adding 1 to the early warning information Tw; and (4) if the absolute value of the residual error in a certain interval does not meet the condition, resetting the Tp and Tw values, returning to the step (1), and restarting to monitor the model prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811011319.9A CN109443419B (en) | 2018-08-31 | 2018-08-31 | Machine learning-based rectifier online monitoring method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811011319.9A CN109443419B (en) | 2018-08-31 | 2018-08-31 | Machine learning-based rectifier online monitoring method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109443419A CN109443419A (en) | 2019-03-08 |
| CN109443419B true CN109443419B (en) | 2021-08-31 |
Family
ID=65530906
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811011319.9A Active CN109443419B (en) | 2018-08-31 | 2018-08-31 | Machine learning-based rectifier online monitoring method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109443419B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111242348B (en) * | 2019-12-30 | 2023-08-01 | 安徽先兆科技有限公司 | Electrical safety monitoring method and system based on time sequence |
| CN111737249A (en) * | 2020-08-24 | 2020-10-02 | 国网浙江省电力有限公司 | Abnormal data detection method and device based on Lasso algorithm |
| CN113009258B (en) * | 2021-03-01 | 2023-10-10 | 上海电气集团数字科技有限公司 | Equipment working state monitoring method |
| CN113468751B (en) * | 2021-07-05 | 2022-12-27 | 河南中烟工业有限责任公司 | Recursion Lasso-based flowmeter anomaly online monitoring method and system and storage medium |
| CN113552855B (en) * | 2021-07-23 | 2023-06-06 | 重庆英科铸数网络科技有限公司 | Industrial equipment dynamic threshold setting method and device, electronic equipment and storage medium |
| CN113609434B (en) * | 2021-08-10 | 2023-08-18 | 中国科学院科技战略咨询研究院 | Method and device for monitoring influence of climate change on forestry |
| CN114236104A (en) * | 2021-10-28 | 2022-03-25 | 阿里云计算有限公司 | Method, device, equipment, medium and product for measuring free calcium oxide |
| CN116667632B (en) * | 2023-06-01 | 2023-11-10 | 广州市科隆节能科技有限公司 | Rectifier control method and control system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0467257A (en) * | 1990-07-09 | 1992-03-03 | Nippon Telegr & Teleph Corp <Ntt> | Neural network circuit |
| CN2128433Y (en) * | 1992-07-23 | 1993-03-17 | 天津市电缆总厂 | Reversible rectifier |
| RU2042195C1 (en) * | 1992-03-12 | 1995-08-20 | Северо-Западный Заочный Политехнический Институт | Device for modeling signals with multiple beams |
| CN2677943Y (en) * | 2003-12-23 | 2005-02-09 | 顺特电气有限公司 | Fault monitoring alarm device for traction rectifier |
| CN103606931A (en) * | 2013-12-10 | 2014-02-26 | 广州供电局有限公司 | AVC (Automatic Voltage Control) system constant value regulation method based on probability statistics characteristic |
| CN105608514A (en) * | 2016-03-31 | 2016-05-25 | 山东大学 | Dynamic heat setting value probability distribution predication method of overhead power transmission line based on quantile regression |
| CN106647285A (en) * | 2017-02-10 | 2017-05-10 | 泉州装备制造研究所 | Catalyst activity detecting method based on soft measurement technology |
| CN106682067A (en) * | 2016-11-08 | 2017-05-17 | 浙江邦盛科技有限公司 | Machine learning anti-fraud monitoring system based on transaction data |
| CN106997509A (en) * | 2017-03-28 | 2017-08-01 | 南京航空航天大学 | A kind of emergency materials distributed needs Forecasting Methodology of uncertain information fusion |
| CN207200595U (en) * | 2017-07-24 | 2018-04-06 | 太仓永瑞检测装备科技有限公司 | Three-level PWM rectifier overall-in-one control schema platform based on DSP and FPGA |
-
2018
- 2018-08-31 CN CN201811011319.9A patent/CN109443419B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0467257A (en) * | 1990-07-09 | 1992-03-03 | Nippon Telegr & Teleph Corp <Ntt> | Neural network circuit |
| RU2042195C1 (en) * | 1992-03-12 | 1995-08-20 | Северо-Западный Заочный Политехнический Институт | Device for modeling signals with multiple beams |
| CN2128433Y (en) * | 1992-07-23 | 1993-03-17 | 天津市电缆总厂 | Reversible rectifier |
| CN2677943Y (en) * | 2003-12-23 | 2005-02-09 | 顺特电气有限公司 | Fault monitoring alarm device for traction rectifier |
| CN103606931A (en) * | 2013-12-10 | 2014-02-26 | 广州供电局有限公司 | AVC (Automatic Voltage Control) system constant value regulation method based on probability statistics characteristic |
| CN105608514A (en) * | 2016-03-31 | 2016-05-25 | 山东大学 | Dynamic heat setting value probability distribution predication method of overhead power transmission line based on quantile regression |
| CN106682067A (en) * | 2016-11-08 | 2017-05-17 | 浙江邦盛科技有限公司 | Machine learning anti-fraud monitoring system based on transaction data |
| CN106647285A (en) * | 2017-02-10 | 2017-05-10 | 泉州装备制造研究所 | Catalyst activity detecting method based on soft measurement technology |
| CN106997509A (en) * | 2017-03-28 | 2017-08-01 | 南京航空航天大学 | A kind of emergency materials distributed needs Forecasting Methodology of uncertain information fusion |
| CN207200595U (en) * | 2017-07-24 | 2018-04-06 | 太仓永瑞检测装备科技有限公司 | Three-level PWM rectifier overall-in-one control schema platform based on DSP and FPGA |
Non-Patent Citations (1)
| Title |
|---|
| 基于小波神经网络的无刷同步发电机旋转整流器故障诊断研究;梁杉等;《四川电力技术》;20080630;第31卷(第3期);第51-53页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109443419A (en) | 2019-03-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109443419B (en) | Machine learning-based rectifier online monitoring method | |
| WO2023197461A1 (en) | Gearbox fault early warning method and system based on working condition similarity evaluation | |
| WO2016101690A1 (en) | Time sequence analysis-based state monitoring data cleaning method for power transmission and transformation device | |
| CN106644162B (en) | Ring main unit wire core temperature soft measurement method based on neighborhood preserving embedding regression algorithm | |
| CN106529719B (en) | Wind power prediction method based on particle swarm optimization algorithm wind speed fusion | |
| CN104595170A (en) | Air compressor monitoring diagnosis system and method adopting adaptive kernel Gaussian hybrid model | |
| CN113255848B (en) | Water turbine cavitation sound signal identification method based on big data learning | |
| CN104537438B (en) | A kind of prediction of peak of power consumption and monitoring method | |
| CN114723285B (en) | Power grid equipment safety evaluation prediction method | |
| CN113036913B (en) | Method and device for monitoring state of comprehensive energy equipment | |
| CN107309213B (en) | Red date ultrasonic cleaner control method based on BP neural network | |
| CN112884089A (en) | Power transformer fault early warning system based on data mining | |
| CN111273125A (en) | RST-CNN-based power cable channel fault diagnosis method | |
| CN114118232A (en) | Intelligent ammeter fault prediction method based on time-space convolution neural network | |
| CN114266301A (en) | Intelligent power equipment fault prediction method based on graph convolution neural network | |
| CN116148753A (en) | Intelligent electric energy meter operation error monitoring system | |
| CN115112978A (en) | Fault detection method for insulating ring main unit of power system | |
| CN116449242A (en) | Battery health state monitoring system for users | |
| CN106325258B (en) | Relay protection device state evaluation method based on online monitoring information | |
| CN105741184B (en) | Transformer state evaluation method and device | |
| CN116187867B (en) | Intelligent transmission module production management method and system | |
| CN115310704A (en) | Multi-dimensional fire safety assessment early warning system based on system safety model | |
| Xu et al. | Study of fault diagnosis based on probabilistic neural network for turbine generator unit | |
| CN114201825A (en) | Method and system for evaluating equipment performance degradation state based on combination characteristics | |
| CN116153007B (en) | Alarm control system and method based on sensor |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200812 Address after: Room 1801, 18 / F, Baiyun electric technology building, No. 9 Keyuan Road, Guangzhou private science and Technology Park, No. 1633 Beitai Road, Baiyun District, Guangzhou City, Guangdong Province Applicant after: Guangzhou Zhixin Power Technology Co., Ltd Applicant after: Baiyun Electric Research Institute (Nanjing) Co., Ltd Applicant after: GUANGZHOU YANGXIN TECHNOLOGY RESEARCH Co.,Ltd. Address before: 510540 Room 105, First Floor, Baiyun Electrical Science and Technology Building, Guangzhou Private Science Park, 1633 Taihe Town, Baiyun District, Guangzhou City, Guangdong Province Applicant before: GUANGZHOU SHIKE HIGH-NEW TECHNOLOGY Co.,Ltd. Applicant before: GUANGZHOU YANGXIN TECHNOLOGY RESEARCH Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |