A method, equipment, and medium for generating characteristic curves of a centrifugal water pump.

Abstract

This invention proposes a method, device, and medium for generating characteristic curves of centrifugal water pumps. The method includes the following steps: S1, constructing a training set; S2, training a discriminator and a generator; S3, performing double-validation denoising; S4, cyclically executing the training and denoising steps until the discriminator and generator fully converge; S5, deriving the QH characteristic curve at the target frequency. This invention can generate characteristic curves of centrifugal water pumps in complex real-world production environments.

Description

Technical Field

[0001] This invention relates to the field of water pump technology, and in particular to a method for generating characteristic curves of centrifugal water pumps based on data augmentation and dual-validation denoising. Background Technology

[0002] Centrifugal pumps, also known as centrifugal water pumps or centrifugal pumps, are pumps that use the centrifugal force generated by the high-speed rotation of an impeller to transport liquids. Their working principle is as follows: when the motor drives the pump shaft and impeller to rotate, water or other liquids are thrown out under the action of centrifugal force, thus forming a low-pressure zone at the center of the impeller. External liquids are forced into this low-pressure zone under atmospheric pressure and then thrown out by the impeller, thus creating a continuous liquid flow. In actual production environments, the operating points of centrifugal pumps at a certain frequency are often concentrated or data for that frequency is unavailable. The distribution defects and scarcity of data make it difficult to fit the curve. Data collected by flow meters and other data acquisition devices will have a certain degree of error; the lower the quality of the data acquisition device, the greater the error. In other words, fluctuations in hardware-acquired data introduce noise data. Noise data under non-steady-state operating conditions, such as when the pump starts and stops, often deviates from its characteristic curve. Summary of the Invention

[0003] This invention aims to at least solve the technical problems existing in the prior art, and in particular, it innovatively proposes a method for generating characteristic curves of centrifugal water pumps based on data augmentation and dual verification denoising.

[0004] To achieve the above-mentioned objectives of this invention, this invention provides a method for generating characteristic curves of a centrifugal water pump based on data augmentation and dual-validation denoising, comprising the following steps:

[0005] Step 1: Construct the training set. Collect real-time operating data of the water pump, calculate the instantaneous head of the water pump, and obtain the actual operating point information of the water pump (frequency, flow rate, head). Based on the data augmentation strategy, generate theoretical equivalent efficiency operating points with frequency variations within ±10% and intervals of 0.1. Use data collected in the most recent period as training data.

[0006] Step 2: Train the discriminator and generator. Using data from three dimensions—frequency, flow rate, and head—as input, and whether the data is real or not as output, train the discriminator of the generative adversarial network (GAN). Using frequency and flow rate as input, and head as output, train the generator of the GAN.

[0007] Step 3, Dual-validation denoising. Set thresholds η1 and η2, and when the generator's loss... G <η1 and the discriminant's loss DWhen η < 2, perform dual-validation denoising, using the discriminator and pump mathematical properties to detect and remove noisy operating points from the training set.

[0008] Step 4: Repeat the training and denoising steps until the discriminator and generator fully converge.

[0009] Step 5: Export the QH characteristic curve at the target frequency. As needed, set the specified frequency and flow range, use the generator to generate the head, obtain a set of (flow rate, head) operating points at the specified frequency, and plot a smooth curve to obtain the QH characteristic curve.

[0010] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above method.

[0011] The present invention also discloses a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above method.

[0012] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are as follows:

[0013] (1) This invention can generate characteristic curves of centrifugal water pumps in complex actual production environments; (2) The generative adversarial framework ensures that the generator can accurately generate operating points that have not appeared in the history when deriving characteristic curves; (3) The data augmentation technology enriches the operating points of the pump operation, effectively improving the generalization ability and accuracy of the model; (4) The dual verification denoising technology effectively eliminates noise data caused by various factors in the actual production environment.

[0014] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0015] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0016] Figure 1 This is a schematic flowchart of the present invention. Detailed Implementation

[0017] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0018] Combination Figure 1 Specifically, it includes the following steps:

[0019] Step 1: Construct the training set. Collect real-time operating data of the water pump, calculate the instantaneous head of the water pump, and obtain the actual operating point information of the water pump (F: frequency, Q: flow rate, H: head). Retain one decimal place for the pump frequency data, and generate theoretical equivalent efficiency operating points with frequency variations within ±10% and intervals of 0.1Hz based on the data augmentation strategy. Use data collected in the most recent period as training data;

[0020] For centrifugal water pumps, the pump efficiency does not change significantly within a certain speed adjustment range. The performance of the centrifugal pump after speed adjustment is as follows: Where n is the original speed of the pump, This refers to the changed rotational speed, while F, Q, and H represent the frequency, flow rate, and head at the original rotational speed. These are the frequency, flow rate, and head after the speed change. Therefore, based on the collected actual operating data, frequency data with intervals of 0.1Hz between [F*0.9, F*1.1] can be generated. And calculate the ratio Therefore, the theoretical operating point (α*F, α*Q, α) with the same efficiency can be calculated. 2 *H). This allows for a significant expansion of the training data with numerous operating conditions.

[0021] The method for calculating the instantaneous head of a water pump is as follows:

[0022]

[0023] Where H represents the instantaneous head of the water pump;

[0024] p represents the pressure of the water pump;

[0025] α represents the pressure coefficient; its value is 102.

[0026] l indicates the installation height of the water pump pressure gauge;

[0027] M represents the liquid level elevation of the suction well;

[0028] Q represents the water pump flow rate;

[0029] β represents the pump flow coefficient; its value is 3600.

[0030] π represents the mathematical constant pi, and its value is 3.14.

[0031] g represents the acceleration due to gravity, and its value is 9.8.

[0032] h represents water loss from the water pump.

[0033] Step 2: Train the discriminator and generator. The discriminator is trained using data from three dimensions: frequency, flow rate, and head, as input and whether the data is real or not as output. The input data is divided into real production data and data generated by the generator, with real data labeled as "true" (1) and data generated by the generator labeled as "false" (0). The generator is trained using frequency and flow rate as input and head as output.

[0034] The entire model framework is divided into two parts: a generator and a discriminator. When training the model using the entire dataset, the discriminator and generator are trained alternately. While training one module, the parameters of the other module are frozen, and the discriminator is trained first. Both the discriminator and generator are three-layer neural networks with 32 nodes in each hidden layer. Specifically, for each set of frequency and flow data (F...) in the database... i Q i Multiplying this by a random noise perturbation coefficient within the range [0.5, 2] yields (F) i ′,Q i The perturbed data is input into the generator G to obtain... Therefore, a new operating point can be obtained through random noise perturbation and a generator. The discriminator takes an operating point as input, and the output layer outputs a number between (0,1) through a sigmoid activation function, representing the probability that the operating point is the pump's true operating point. When training the discriminator, the loss function is:

[0035]

[0036] Where G(·) represents the generator mapping function and D(·) represents the discriminator mapping function.

[0037] N represents the number of operating points;

[0038] F i Q i H i These are the frequency, flow rate, and head at the original speed at the i-th operating point, respectively.

[0039] F i ′,Q i ′ represent the frequency and flow rate after the speed change at the i-th operating point, respectively;

[0040] When training the generator, the loss function is:

[0041]

[0042] Step 3, Dual-validation denoising. Set thresholds η1 and η2, and when the generator's loss... G<η1 and the discriminator loss is 0.5-η2 <loss D When <0.5+η2, perform double validation denoising: use a discriminator to obtain the set S1={(F i Q i H i )|i=1,2,…,N;D(F i Q i H i <0.4}; In addition, the least squares method is used to fit the quadratic equation H=aQ of the relationship between flow rate and head at the operating points of the same frequency. 2 +bQ+c, marking the set S2 of operating points whose variance with the fitted curve is greater than the threshold σ. This yields the intersection S = S1∩S2 of operating points that neither satisfy the discriminator's criteria nor the mathematical laws governing the mechanism. The set S of operating points is then removed from the training dataset to reduce noisy data in the training set.

[0043] Step 4: Repeat the training and denoising steps until the discriminator and generator fully converge.

[0044] Observe the trends of the loss functions of the generator (G) and discriminator (D) as training iterations progress. Ideally, the generator's loss should gradually decrease, while the discriminator's loss should approach 0.5. Training ends when the generator's loss can no longer decrease and the discriminator's loss approaches 0.5.

[0045] Step 5: Export the QH characteristic curve at the target frequency. After training, extract the generator G, select a frequency F as needed, and within the selected flow range [Q min Q max The water pump generates a uniform flow rate within the pump. M represents the selected flow rate Q i Quantity, Q min To Q max The denser the intervals, the larger M becomes. The input generator obtains the head. Therefore, a set of flow rate and head operating points at frequency F can be plotted on the coordinate axis. Plot a smooth curve to obtain the QH characteristic curve of the pump at frequency F.

[0046] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for generating characteristic curve of centrifugal water pump based on data augmentation and double verification denoising, characterized in that, The method comprises the following steps: S1, constructing a training set; S2, training a discriminator and a generator; the discriminator and the generator are trained alternately, the parameters of one module are frozen when training the other module, and the discriminator is trained first; the discriminator and the generator are both three-layer neural networks with 32 nodes in each hidden layer; When training the discriminator, the loss function is: Wherein G(·) represents a generator mapping function, D(·) represents a discriminator mapping function, N represents the number of operating points; F i ,Q i ,H i respectively the frequency, flow rate and head at the original rotational speed of the i-th operating point. F i ′ ,Q i ′ respectively, the frequency and flow rate after the speed of the i th operating point is changed When training the generator, the loss function is S3, double verification denoising; S4, repeatedly performing the training step and the denoising step until the discriminator and the generator are completely converged; S5, deriving a QH characteristic curve at a target frequency.

2. The method for generating characteristic curve of centrifugal water pump based on data augmentation and double verification denoising according to claim 1, characterized in that, In step S1, the method for constructing the training set is: Real-time operation data of the water pump is collected, the instantaneous lift of the water pump is calculated, and the operating point information of the actual operation of the water pump is obtained.

3. The method for generating characteristic curve of centrifugal water pump based on data augmentation and double verification denoising according to claim 2, characterized in that, The real-time operation data of the water pump includes frequency and flow rate; The operating point information of the actual operation of the water pump includes frequency, flow rate and instantaneous lift.

4. The method for generating characteristic curve of centrifugal water pump based on data augmentation and double verification denoising according to claim 1, characterized in that, In step S2, the method for training the discriminator and the generator is: The data in three dimensions of frequency, flow rate and instantaneous lift are taken as input, and whether it is real data is taken as output, to train the discriminator of the generative adversarial network; The frequency and flow rate are taken as input, and the instantaneous lift is taken as output, to train the generator of the generative adversarial network.

5. The method for generating characteristic curve of centrifugal water pump based on data augmentation and dual verification denoising according to claim 1, characterized in that, In step S3, the method for double verification denoising is: Set threshold η1 and η2, when the loss loss G <η1 of the generator and the loss loss D <η2 of the discriminator, perform double verification denoising, detect noise working points from the training set and delete using the discriminator and pump mathematical properties.

6. The method for generating characteristic curve of centrifugal water pump based on data augmentation and dual verification denoising according to claim 1, characterized in that, In step S5, the method for deriving a QH characteristic curve at a target frequency is: A specified frequency and flow rate range are set, the generator is used to generate lift, a group of operating points at the specified frequency are obtained, a smooth curve is drawn to obtain a QH characteristic curve.

7. A computer device, comprising: A computer program product, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method of any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, A computer program product, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method of any one of claims 1-6.

Images