阅读说明：本技术 一种基于效能分析的水泵运行状态异常判别的方法 (Water pump running state abnormity discrimination method based on efficiency analysis ) 是由 陶清宝 汤俊萱 肖雷 鲍劲松 于 2020-04-28 设计创作，主要内容包括：本发明公开了一种基于效能分析的水泵运行状态异常判别的方法,根据收集到的水泵运行状态中的电流和轴功率数据,分析了电流和轴功率相关关系,将电流和轴功率之间的关系转换为流量和效率之间的关系,从而根据水泵效能曲线来说明电流和效率这之间可能存在的关系,随后根据已收集到的数据进行多项式回归,并进行统计分析,最后通过残差正态分布3σ法则判断水泵运行是否存在异常,据此判断水泵是否存在潜在的故障。(The invention discloses a water pump running state abnormity discrimination method based on efficiency analysis, which analyzes the correlation between current and shaft power according to the collected current and shaft power data in the water pump running state, converts the relation between the current and the shaft power into the relation between flow and efficiency, thereby explaining the possible relation between the current and the efficiency according to a water pump efficiency curve, then carries out polynomial regression according to the collected data, carries out statistical analysis, and finally judges whether the water pump runs abnormally or not according to a residual normal distribution 3 sigma rule, thereby judging whether the water pump has potential faults or not.)

1. A method for judging the abnormal operation state of a water pump based on efficiency analysis is characterized by comprising the following steps:

step S1: acquiring two state data of instantaneous current and water pump shaft power at corresponding moment in the operation process of the water pump by a sensor arranged on the water pump;

step S2: analyzing the mechanism of the water pump to obtain a performance curve fitting type;

step S3: fitting the collected state monitoring data with a polynomial;

step S4: and (5) carrying out analysis detection test on abnormal points of the running state by using a normal distribution 3 sigma criterion.

2. The method for determining the abnormal operating condition of the water pump based on the performance analysis of claim 1, wherein the step S2 includes the following steps:

step S21: calculating the output power of the water pump motor under the instantaneous current in the step S1, and calculating the formulaP is the output power of the motor, U is the line voltage, I represents the current currently flowing through the motor,is a power factor ofDetermining a load type;

step S22: the efficiency of the water pump reaches the maximum value when it is close to the rated power, and the efficiency is low when it is lower than the rated power and higher than the rated power, and the formula in step S21 is rewritten into the formula after considering the efficiency of the water pump itselfP_aRepresenting the shaft power of the water pump, U is the line voltage, I is the current currently flowing through the motor,for power factor, η represents the efficiency of the water pump itself;

step S23: the total efficiency of the water pump is calculated as a dependent variable according to the equation on the left side of the equation in the step S22 with the motor current as an independent variable, so that the distribution of the water pump efficiency η -current I of the water pump motor at different moments can be obtained.

3. The method for determining the abnormal operating condition of the water pump based on the performance analysis of claim 2, wherein U in the step S21 is 380V when the industrial power is used,is 0.8.

4. The method for determining the abnormal operating condition of the water pump based on the performance analysis of claim 1, wherein the step S3 includes:

step S31: carrying out cubic polynomial fitting on the data by using a Matlab tool box, wherein the fitting criterion is least square fitting, and the calculation target of the least square method is

step S32: calculating a regression statistical index decision coefficient R²And the sample standard deviation sigma of the residual error is calculated by the formulaR²Representing the decision coefficient, y representing the actual value of the sample,represents the value of the regression function,represents the average of the sample values, σ represents the sample standard deviation of the residual, and n represents the number of samples.

5. The method for determining abnormality in operating condition of water pump based on performance analysis of claim 4, wherein the random error of the polynomial linear regression in step S4 satisfies N (0, σ)²) The distribution, i.e. the normal distribution with a mean value of 0 and a standard deviation of σ, can be tested according to the 3 σ rule of the normal distribution, including the abnormal values in the training samples and the abnormal values in the newly input data, and the test is performed on the originally input data after the regression curve is established, if the difference value from the fitting curve exceeds 3 σ, the data is considered to belong to the abnormal point outside the confidence interval of the random error, and then the abnormal point needs to be removed and the fitting is performed again until the abnormal point does not exist in the training data, and at this time, the establishment of the regression model is completed.

6. The method as claimed in claim 5, wherein after the regression model is built in step S4, if there is new data, the model is input for detection, and if the difference between the new data and the regression value is greater than 3 σ, it is determined that the difference is not caused by random error but is abnormal.

Technical Field

The invention relates to a water pump running state abnormity discrimination technology based on efficiency analysis, in particular to a water pump running state abnormity discrimination method based on efficiency analysis.

Background

Water pumps have a wide and important application in industrial production. Once the water pump breaks down, not only can lead to equipment self to stop working, production processes around more can influencing makes enterprise's profit reduce by a wide margin. Existing enterprises typically repair equipment with post-corrective repairs or periodic preventative maintenance, which may perpetuate failures or incur unnecessary maintenance costs. If can just judge the future trouble that probably takes place of analysis water pump according to the state information of water pump before the water pump trouble takes place to maintain in advance and maintain in advance, then just so can reduce down time by a wide margin, reduce the dead time, thereby guarantee the normal production of whole production line, thereby guarantee the profit of enterprise.

Disclosure of Invention

In order to solve the problems, the invention provides a method for judging the abnormal running state of a water pump based on efficiency analysis, which analyzes the correlation between current and shaft power according to the collected data of the current and the shaft power in the running state of the water pump, converts the relationship between the current and the shaft power into the relationship between flow and efficiency, thereby explaining the possible relationship between the current and the efficiency according to a water pump efficiency curve, then performs polynomial regression according to the collected data, performs statistical analysis, and finally judges whether the water pump runs abnormally or not according to a residual normal distribution 3 sigma rule so as to judge whether the water pump has potential faults or not.

The invention relates to a method for judging the abnormal running state of a water pump based on efficiency analysis, which comprises the following steps:

step S1: acquiring two state data of instantaneous current and water pump shaft power at corresponding moment in the operation process of the water pump by a sensor arranged on the water pump;

step S2: analyzing the mechanism of the water pump to obtain a performance curve fitting type;

step S3: fitting the collected state monitoring data with a polynomial;

step S4: and (5) carrying out analysis detection test on abnormal points of the running state by using a normal distribution 3 sigma criterion.

In the foregoing solution, the step S2 includes the following steps:

step S21: calculating the output power of the water pump motor under the instantaneous current in the step S1, and calculating the formulaP is the output power of the motor, U is the line voltage, I represents the current currently flowing through the motor,is a power factor, determined by the load type;

step S22: the efficiency of the water pump reaches the maximum value when it is close to the rated power, and the efficiency is low when it is lower than the rated power and higher than the rated power, and the formula in step S21 is rewritten into the formula after considering the efficiency of the water pump itself

P_aRepresenting the shaft power of the water pump, U is the line voltage, I is the current currently flowing through the motor,

for power factor, η represents the efficiency of the water pump itself;

step S23: the total efficiency of the water pump is calculated as a dependent variable according to the equation on the left side of the equation in the step S22 with the motor current as an independent variable, so that the distribution of the water pump efficiency η -current I of the water pump motor at different moments can be obtained.

In the scheme, U in the step S21 is 380V when the industrial power is used,

is 0.8.

In the foregoing solution, the step S3 includes:

step S31: carrying out cubic polynomial fitting on the data by using a Matlab tool box, wherein the fitting criterion is least square fitting, and the calculation target of the least square method isN is the number of samples, f (x)_i) Is the regression value of the fitted function, y_iIs the actual value, finding the cubic curve that minimizes the sum of the squared residuals;

step S32: calculating a regression statistical index decision coefficient R²And the sample standard deviation sigma of the residual error is calculated by the formula

R²Representing the decision coefficient, y representing the actual value of the sample,represents the value of the regression function,

represents the average of the sample values, σ represents the sample standard deviation of the residual, and n represents the number of samples.

In the above scheme, the random error of the polynomial linear regression in the step S4 satisfies N (0, σ)²) The distribution, i.e. the normal distribution with a mean value of 0 and a standard deviation of σ, can be tested according to the 3 σ rule of the normal distribution, including the abnormal values in the training samples and the abnormal values in the newly input data, and the test is performed on the originally input data after the regression curve is established, if the difference value from the fitting curve exceeds 3 σ, the data is considered to belong to the abnormal point outside the confidence interval of the random error, and then the abnormal point needs to be removed and the fitting is performed again until the abnormal point does not exist in the training data, and at this time, the establishment of the regression model is completed.

In the above scheme, after the regression model is built in step S4, if there is new data, the model is input for detection, and if the difference between the new data and the regression value is greater than 3 σ, it is considered that the difference is not caused by a random error, but is an abnormal point.

The invention has the advantages and beneficial effects that:

1. from the aspect of water pump efficiency analysis, the method analyzes the internal relation between current and power of two parameters, finds a performance curve of the water pump as a bridge in the middle of the fitting process, and further determines the fitting method and the times of fitting a polynomial;

2. according to the invention, an abnormality detection model of the water pump can be established from sample data with less data dimension and less data quantity, and meanwhile, the problem of less fault data quantity in the collected sample can be solved;

3. the invention can establish an abnormality detection model corresponding to each water pump one by one;

4. the method can quickly and conveniently establish the abnormal detection model of the water pump running condition, and has strong engineering practice capability.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 is a polynomial fit of a water pump performance curve in an embodiment.

FIG. 2 is a flow chart of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.

As shown in fig. 1 and 2, the present invention is a method for determining abnormal operating conditions of a water pump based on performance analysis, including the following steps:

step S1: acquiring two state data of instantaneous current and water pump shaft power at corresponding moment in the operation process of a water pump by a sensor arranged on a water pump motor;

the data of the motor current and the water pump shaft power of the water pump in the actual working process of a certain energy station are collected by the specific embodiment, the application types of the water pump comprise a primary pump, a secondary pump and the like, and the state data of the running time of the secondary pump is collected by taking a secondary pump No. 02 cold supply pump as an example.

Step S2: analyzing the mechanism of the water pump to obtain a performance curve fitting type;

calculating the output power of the water pump motor under the instantaneous current in the step S1 by the formula

Wherein P is the output power of the motor, U is the line voltage, 380V is taken when the industrial power is used, I represents the current flowing through the motor,

is the power factor, is determined by the load type, and can generally take an empirical value of 0.8 for a water pump motor used in engineering practice.

Theoretically, the shaft power of the water pump should be proportional to the current, but the efficiency of the water pump is different in different operating conditions. From the water pump performance curve, the efficiency of the water pump reaches a maximum when it is near the rated power, and is less efficient below and above the rated power.

The performance curves of most water pumps are not exactly the same, but the trends of the curves are consistent. Therefore, the formula (1) can be rewritten into the formula (1) in consideration of the efficiency of the water pump itself

Wherein, P_aRepresenting the shaft power of the water pump, U is the line voltage, I is the current currently flowing through the motor,for power factor, η represents the efficiency of the water pump itself.

When the motor current is larger, the power of the water pump is larger, and the water pump can pump more liquid out to do work, so that positive correlation exists between the flow and the motor current. In this case, the variation trend of the efficiency η -flow Q curve, which is one of the performance curves of the water pump, can also be considered as the variation trend of the efficiency η -current I. Therefore, if the motor current is used as an independent variable, and the total efficiency of the water pump is calculated as a dependent variable according to the equation on the left side of the equation (2), the distribution situation of the water pump efficiency eta-current I of the water pump motor at different moments can be obtained.

Meanwhile, in the literature "analytical expression method of water pump performance curve" of Zhao Linming in journal "irrigation and Small hydropower", authors successfully fit the performance curves of two types of water pumps by using cubic polynomial, and find that fitting the performance curve of the water pump by using cubic curve, especially the relation between efficiency and flow rate, can obtain better effect. Therefore, according to the characteristic that the efficiency in the performance curve is in positive correlation with the process, the data points are also fitted by using the cubic polynomial.

Step S3: fitting the collected state monitoring data with a polynomial;

and (3) carrying out cubic polynomial fitting on the data by using a Matlab tool box, wherein the fitting criterion is least square fitting, and the calculation target of the least square method is as follows:

wherein N is the number of samples, f (x)_i) Is the regression value of the fitted function, y_iIs the actual value, finding the cubic curve that minimizes the sum of the squared residuals;

then calculating the regression statistical index decision coefficient R²And the sample standard deviation sigma of the residual error is calculated by the formula

In the formula, R²Representing the decision coefficient, y representing the actual value of the sample,represents the value of the regression function,represents the average of the sample values, σ represents the sample standard deviation of the residual, and n represents the number of samples.

Fitting by using a Matlab tool box, wherein the fitting result is shown in figure 1, the vertical axis represents the overall efficiency multiplied by the efficiency of each link of the water pump, the horizontal axis represents the current of the motor, the point with the difference value exceeding 3 sigma of the fitting curve is taken as an abnormal point deletion cycle for retraining according to the judgment criterion of 3 sigma, and after all training is finished (the regression fitting is performed twice in the invention), the polynomial finally obtained by model fitting is-5.013 × 10^-8x³+3.382×10^-5x²+0.0094x-0.6216, sample standard deviation σ of the residual is 0.0463, determining the coefficient R²0.9707, this represents that the regression model can represent 97% of the data changes.

Step S4: carrying out running state abnormal point analysis detection test by using a normal distribution 3 sigma criterion;

theoretically, the water pump efficiency calculated by the formula (2) according to actually measured data should fall on an efficiency regression curve of the water pump. However, in the actual production process, due to the factors of water pump equipment faults and measurement inaccuracy such as unstable voltage, damaged bearings, rotor dynamic unbalance and the like, the calculated efficiency deviates from an ideal regression curve. Because the efficiency of the water pump is calculated by current and voltage values, the excessive fluctuation of the efficiency can be considered to be caused by the mismatching of current and power values, and at the moment, the water pump equipment is always in an abnormal operation condition or a fault occurs at a certain part of the equipment, so that whether the fluctuation exceeds a specified range or not needs to be judged by calculation.

The random error of the polynomial linear regression satisfies N (0, sigma) obtained from the mathematical statistical knowledge²) Distribution, i.e. normal distribution with mean 0 and standard deviation σ, and the error here can be considered to be due to random, countless, independent, multiple factors, so outliers can be tested according to the 3 σ -law of normal distribution, including outliers in training samples and in new input dataAn outlier. After the regression curve is established, the original input data is checked, and if the difference value with the fitted curve exceeds 3 sigma, the data is considered to belong to an abnormal point outside the confidence interval of random error. Outliers then need to be removed and the fitting repeated until there are no outliers in the training data. At this point, a regression model is established, and if new data is available, the model can be input for detection, and if the difference value with the regression value is more than 3 sigma, the model is not considered to be caused by random errors, but is considered to be an abnormal point. In the invention, the abnormal point may represent that the equipment has a fault, and may be caused by non-fault reasons such as an overlarge starting current when the water pump motor is started.

And after new test data are input again, judging according to the 3 sigma criterion, if the data are considered as abnormal points by the model, outputting 1, and if the data are considered to be in a normal range, outputting 0. In the model, 7 data at new moments are input for testing, and the obtained result is that 6 data are all in a normal range, and one data is considered as an abnormal point by the regression model, which may be caused by the existence of faults of equipment.

According to the collected current and shaft power data in the water pump running state, the invention analyzes the correlation between the current and the shaft power, converts the relation between the current and the shaft power into the relation between flow and efficiency, and explains the possible relation between the current and the efficiency according to the water pump efficiency curve. Polynomial regression was then performed on the collected data and statistical analysis was performed. And finally, judging whether the water pump runs abnormally or not through a residual normal distribution 3 sigma rule, and judging whether the water pump has potential faults or not according to the judgment. The advantages are that:

1. from the aspect of water pump efficiency analysis, the method analyzes the internal relation between current and power of two parameters, finds a performance curve of the water pump as a bridge in the middle of the fitting process, and further determines the fitting method and the times of fitting a polynomial;

2. according to the invention, an abnormality detection model of the water pump can be established from sample data with less data dimension and less data quantity, and meanwhile, the problem of less fault data quantity in the collected sample can be solved;

3. the invention can establish an abnormality detection model corresponding to each water pump one by one;

4. the method can quickly and conveniently establish the abnormal detection model of the water pump running condition, and has strong engineering practice capability.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.