Control method and control system of lift pump

Abstract

The invention relates to the technical field of lift pump regulation and control, in particular to a lift pump control method and a lift pump control system, and overcomes the defects of electricity consumption and operation cost waste caused by the fact that a lift pump cannot keep efficient operation all the time in the prior art. Calculating the optimal lifting water quantity under the liquid level of the pump room, and then training an SSA-SVR prediction model; after the model predicts the optimal lifting water quantity according to the new pump room liquid level and the water inlet and outlet water quality parameters, the number of the started lifting pumps is calculated according to the optimal lifting water quantity, the rated parameters of the lifting pumps and the optimal operation frequency, and finally the operation of the lifting pumps is regulated and controlled according to the optimal operation frequency and the number of the started lifting pumps, so that the energy utilization efficiency can be optimized, and the energy utilization rate is improved. And intelligent and refined operation management of the lifting pump is realized. In addition, different actual working conditions can be dealt with, the lifting pump is kept to operate efficiently, and waste of power consumption cost and labor cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of lift pump regulation, and more specifically, to a control method and control system for a lift pump. Background Art

[0002] In the treatment process of a water purification plant, water needs to pass through multiple treatment units (such as sedimentation, filtration, disinfection, etc.). The lift pumps in the pump house provide power for the water flow, lifting the water from a lower position to a higher position to ensure that the water can smoothly pass through each treatment link to meet the process requirements in the water purification process and ensure the stable operation of the entire water purification plant.

[0003] The high power consumption characteristic makes the lift pump one of the main energy-consuming devices in the water purification plant, and its annual power consumption usually accounts for more than 20% of the total power consumption of the sewage treatment plant. Due to certain deviations between the actual operating conditions of some plants and the preset operating conditions, the lift pump cannot operate efficiently at all times, and manual regulation is unstable, which also causes the lift pump to be unable to maintain efficient operation, ultimately resulting in waste of power consumption and operating costs. Summary of the Invention

[0004] Aiming at the problem of waste of power consumption and operating costs caused by the inability of the lift pump to always maintain efficient operation in the above-mentioned prior art, the present invention provides a control method and control system for a lift pump, which can make the lift pump always maintain efficient operation and reduce the waste of power consumption and operating costs.

[0005] To solve the above technical problems, the technical solution provided by the present invention is:

[0006] A control method for a lift pump, comprising the following steps:

[0007] S1: Collect historical data of multiple groups of pump house liquid levels, influent water quality parameters, and effluent water quality parameters.

[0008] S2: Calculate the optimal lifting water volume at the pump house liquid level to obtain a data sample set.

[0009] S3: Use the pump house liquid level, influent water quality parameters, and effluent water quality parameters in the data sample set as input features, and use the optimal lifting water volume as the output feature to train the SSA-SVR prediction model.

[0010] S4: The SSA-SVR prediction model predicts the corresponding optimal lifting water volume prediction value according to the new pump house liquid level, influent water quality parameter, and effluent water quality parameter data.

[0011] S5: Calculate the optimal operating frequency of the lift pump corresponding to the new pump house liquid level; then calculate the number of lift pumps to be started according to the optimal operating frequency, the optimal lifting water volume prediction value, and the rated flow of the lift pump.

[0012] S6: Regulate and control the lift pump according to the optimal operating frequency and the number of activated units.

[0013] Preferably, in the step S1, both the influent water quality parameters and the effluent water quality parameters include the chemical oxygen demand concentration, ammonia nitrogen concentration, and suspended solid concentration.

[0014] Preferably, in the step S3, the training steps of the SSA-SVR prediction model are as follows:

[0015] S31: Divide the data sample set into a training set and a test set;

[0016] S32: Use the pump house liquid level, influent water quality parameters, and effluent water quality parameters in the data sample set as input features, and use the optimal lift water volume as the output feature; Initialize the population using the Sparrow Search Algorithm (SSA), calculate the mean square error of each sparrow individual through the SVR model; Update the sparrow population position according to the behaviors of the discoverers, followers, and scouts of the sparrows; Then repeat the position update and mean square error calculation until the maximum number of iterations is reached or the mean square error value converges; Finally, perform cross-validation using GridSearchCV that defines the hyperparameter range near the approximate optimal solution found by SSA to find the global optimal solution, and use the global optimal solution as the optimal hyperparameters of the SVR model;

[0017] S33: Retrain the SVR model using the optimal hyperparameters;

[0018] S34: Calculate the mean square error, mean absolute error, and determination coefficient of the model based on the test set to evaluate the performance of the model. If the model performance meets the requirements, select it as the final SSA-SVR prediction model; otherwise, adjust the model parameters and return to step S31 again.

[0019] Preferably, in the step S32, the SVR model uses the mean square error function to measure the fitness, and the specific algorithm is as follows:

[0020]

[0021] where y i is the true value; is the predicted value.

[0022] Preferably, in the step S32, the SSA algorithm imitates the habits of sparrows, and uses three formulas to define the discoverers, followers, and scouts respectively, combining the effects of global search, local search, and regularization optimization to optimize the hyperparameters in the SVR. Among them, the defining formula of the discoverer is:

[0023]

[0024] X i = X i + Q * L for other cases;

[0025] The definition formula of the follower is:

[0026]

[0027] X i = X p + |X i - X p | * A + * L for other cases;

[0028] The definition formula of the vigilant is:

[0029] X i = X best + β|X best - X i |;

[0030] In the formula, X i is the position of the i-th sparrow in the iteration; X best is the position of the current optimal individual; T is the maximum number of iterations; R is a random number with a value range of [0, 1], used to determine whether to update the position; ST is the safety threshold with a value range of [0.5, 1], used to determine whether to update the position; Q is a random number subject to the standard normal distribution N(0, 1), used to adjust the amplitude of position update; L is the identity matrix, used to adjust the direction of position update; n is the total number of sparrows; X p is the position of the discoverer in the previous iteration; A + is a random matrix with element value range of [-1, 1], used to adjust the direction of position update; X worst is the position of the worst individual; α and β are both step control parameters, which are random numbers subject to the normal distribution with a mean of 0 and a variance of 1.

[0031] Preferably, in the step S5, the calculation formula of the optimal operating frequency is:

[0032]

[0033] In the formula, f is the optimal operating frequency of the lift pump, Hz; H is the total head, m; H1 is the rated head of the lift pump, m; where, the calculation formula of the total head is:

[0034] H = H2 - H3 + Hfa;

[0035] In the formula, H2 is the pump house lifting height, m; H3 is the pump house liquid level, m; H fais the actual pipeline hydraulic loss; wherein, the calculation formula of the actual pipeline hydraulic loss is:

[0036]

[0037] In the formula, Q1 is the predicted value of the optimal lifting water volume, m 3 / h; Q0 is the rated flow of the lifting pump, m3 / h; H f is the designed hydraulic loss, m; wherein, the calculation formula of the designed hydraulic loss is:

[0038] Hf = H1 - H2 + H4;

[0039] In the formula, H f is the designed hydraulic loss, m; H1 is the rated head of the lifting pump, m; H2 is the lifting height of the pump house, m; H4 is the designed liquid level of the pump house, m. Correcting the designed hydraulic loss to obtain the actual pipeline hydraulic loss can improve the accuracy of the data.

[0040] Preferably, in the step S5, the calculation formula of the number of operating units is:

[0041]

[0042] In the formula, f a is the optimal operating frequency of the lifting pump, Hz; Q1 is the predicted value of the optimal lifting water volume, m 3 / h; Q0 is the rated flow of the lifting pump, m 3 / h.

[0043] Preferably, before executing the step S3, first eliminate the outliers in each column of the data sample set.

[0044] Preferably, the interquartile range (IQR) criterion is used to eliminate the outliers. The specific steps are as follows: calculate the first quartile and the third quartile for each column of data, and obtain the interquartile range according to the difference between the two; then calculate the upper and lower bounds of the outliers according to the first quartile, the third quartile and the interquartile range; finally, eliminate the outliers less than the lower bound and greater than the upper bound in each column of data from the data set.

[0045] The present invention also provides a control system for a lift pump, which is used to execute the above control method, and includes: a data acquisition module for collecting data; a hydraulic parameter calculation module, which is signal-connected to the data acquisition module, and is used to receive the historical data collected by the data acquisition module and calculate the optimal operating frequency; a data optimization module, which is respectively signal-connected to the data acquisition module and the hydraulic parameter calculation module, and is used to eliminate outliers in the data; an SSA-SVR prediction module, which is respectively signal-connected to the data acquisition module, the hydraulic parameter calculation module, and the data optimization module, and is used to receive data to train the model and predict the optimal lift water volume; a decision module, which is signal-connected to the SSA-SVR prediction module, and is used to receive the data of the optimal lift water volume prediction value and calculate the corresponding number of lift pumps to be started; an equipment control module, which is respectively signal-connected to the SSA-SVR prediction module and the decision module, and is used to be signal-connected to the lift pump and control the number of lift pumps to be started and the operating frequency.

[0046] Advantages of the present invention: By collecting the historical data of the pump house liquid level, inlet water quality parameters, and outlet water quality parameters, and calculating the optimal lift water volume under the pump house liquid level, the SSA-SVR prediction model is trained using the foregoing data; after the model predicts the optimal lift water volume according to the new pump house liquid level, inlet water quality parameters, and outlet water quality parameters, the number of lift pumps to be started is calculated according to the optimal lift water volume, the rated parameters of the lift pump, and the optimal operating frequency. Finally, the operation of the lift pump is regulated according to the optimal operating frequency and the number of lift pumps to be started, which can optimize the energy utilization efficiency and realize the intelligent and refined operation management of the lift pump. In addition, the present invention can cope with different actual working conditions, keep the lift pump running efficiently, and reduce the waste of power consumption cost and labor cost. Description of the Drawings

[0047] Figure 1 is a flowchart of a control method for a lift pump;

[0048] Figure 2 is a training flowchart of the SSA-SVR prediction model. Detailed Embodiments

[0049] The following will further specifically describe the technical solutions of the present invention through specific embodiments in conjunction with the drawings:

[0050] Embodiment 1

[0051] This embodiment is the first embodiment of a control method for a lift pump. As Figure 1 shown, it includes the following steps:

[0052] S1: Collect multiple groups of historical data of the pump house liquid level, inlet water quality parameters, and outlet water quality parameters.

[0053] S2: Calculate the optimal lifting water volume under the pump house liquid level to obtain a data sample set.

[0054] S3: Use the pump house liquid level, influent water quality parameters, and effluent water quality parameters in the data sample set as input features, and the optimal lifting water volume as the output feature to train the SSA-SVR prediction model.

[0055] S4: The SSA-SVR prediction model predicts the corresponding optimal lifting water volume prediction value according to the new pump house liquid level, influent water quality parameters, and effluent water quality parameter data.

[0056] S5: Calculate the optimal operating frequency of the lift pump corresponding to the new pump house liquid level; then calculate the number of lift pumps to be started according to the optimal operating frequency, the optimal lifting water volume prediction value, and the rated flow of the lift pump.

[0057] S6: Regulate the lift pump according to the optimal operating frequency and the number of pumps to be started.

[0058] Further, in step S1, the influent water quality parameters and the effluent water quality parameters both include the chemical oxygen demand concentration, ammonia nitrogen concentration, and suspended solid concentration.

[0059] Beneficial effects of this embodiment: By collecting historical data of the pump house liquid level, influent and effluent water quality parameters, calculating the optimal lifting water volume under the pump house liquid level, and then training the SSA-SVR prediction model; after the model predicts the optimal lifting water volume according to the new pump house liquid level, influent and effluent water quality parameters, calculate the number of lift pumps to be started according to the optimal lifting water volume, the rated parameters of the lift pump, and the optimal operating frequency, and finally regulate the operation of the lift pump according to the optimal operating frequency and the number of pumps to be started, the energy utilization efficiency can be optimized, and the intelligent and refined operation management of the lift pump can be realized. In addition, the present invention can cope with different actual working conditions, keep the lift pump running efficiently, and reduce the waste of power consumption cost and labor cost.

[0060] Embodiment 2

[0061] This embodiment is the second embodiment of a control method for a lift pump. On the basis of Embodiment 1, steps S3 to S6 are further optimized.

[0062] Further, as shown in Figure 1 and Figure 2 In step S3, the training steps of the SSA-SVR prediction model are:

[0063] S31: Divide the data sample set into an 80% training set and a 20% test set;

[0064] S32: Use the pump house liquid level, influent water quality parameters, and effluent water quality parameters in the data sample set as input features, and use the optimal water lifting volume as the output feature (i.e., the label); initialize the population using the Sparrow Search Algorithm (SSA), calculate the mean square error of each sparrow individual through the SVR model; update the sparrow population position according to the behaviors of the discoverers, followers, and vigilant sparrows; then repeat the position update and mean square error calculation until the maximum number of iterations is reached or the mean square error value converges; finally, perform cross-validation using GridSearchCV that defines the hyperparameter range near the approximate optimal solution found by SSA to find the global optimal solution, and use the global optimal solution as the optimal hyperparameters of the SVR model;

[0065] S33: Retrain the SVR model using the optimal hyperparameters;

[0066] S34: Calculate the mean square error, mean absolute error, and coefficient of determination of the model based on the test set to evaluate the performance of the model. If the model performance meets the requirements, select it as the final SSA-SVR prediction model; otherwise, adjust the model parameters and return to step S31 again.

[0067] Furthermore, in step S32, the SVR model uses the mean square error function to measure fitness, and the specific algorithm is as follows:

[0068]

[0069] where yi is the true value; is the predicted value.

[0070] Furthermore, in step S32, the SSA algorithm imitates the habits of sparrows and uses three formulas to define the discoverers, followers, and vigilant sparrows respectively, combining global search, local search, and regularization optimization effects to optimize the hyperparameters in SVR. Among them, the definition formula of the discoverer is:

[0071]

[0072] X i = X i + Q * L for other cases;

[0073] The definition formula of the follower is:

[0074]

[0075] X i = X p + |X i - X p | * A + * L for other cases;

[0076] The definition formula of the vigilant sparrow is:

[0077] X i = X best + β|X best - X i |;

[0078] In the formula, X i is the position of the i-th sparrow in the iteration; X best is the position of the current optimal individual; T is the maximum number of iterations; R is a random number with a value range of [0, 1], used to determine whether to update the position; ST is a safety threshold with a value range of [0.5, 1], used to determine whether to update the position; Q is a random number that follows the standard normal distribution N(0, 1), used to adjust the amplitude of position update; L is an identity matrix, used to adjust the direction of position update; n is the total number of sparrows; X p is the position of the discoverer in the previous iteration; A + is a random matrix with element value range of [-1, 1], used to adjust the direction of position update; X worst is the position of the worst individual; α and β are both step size control parameters, which are random numbers that follow the normal distribution with a mean of 0 and a variance of 1.

[0079] Preferably, in step S32, the SSA algorithm imitates the habits of sparrows and uses three formulas to define the discoverer, follower, and vigilant, respectively, to combine the effects of global search, local search, and regular optimization to optimize the hyperparameters in SVR. Among them, the definition formula of the discoverer is:

[0080]

[0081] X i = X i + Q * L for other cases;

[0082] The definition formula of the follower is:

[0083]

[0084] X i = X p + |X i - X p | * A + * L for other cases;

[0085] The definition formula of the vigilant is:

[0086] X i = X best + β|X best - X i |;

[0087] In the formula, X iThe position of the $i$-th sparrow in the iteration; $X$ best is the position of the current optimal individual; $T$ is the maximum number of iterations; $R$ is a random number with a value range of $[0, 1]$, used to determine whether to update the position; $ST$ is a safety threshold with a value range of $[0.5, 1]$, used to determine whether to update the position; $Q$ is a random number following the standard normal distribution $N(0, 1)$, used to adjust the amplitude of position update; $L$ is the identity matrix, used to adjust the direction of position update; $n$ is the total number of sparrows; $X$ p is the position of the discoverer in the previous iteration; $A$ + is a random matrix with element value range of $[-1, 1]$, used to adjust the direction of position update; $X$ worst is the position of the worst individual; both $\alpha$ and $\beta$ are step size control parameters, which are random numbers following the normal distribution with a mean of 0 and a variance of 1.

[0088] Furthermore, in step S5, the calculation formula for the optimal operating frequency is:

[0089]

[0090] In the formula, $f$ is the optimal operating frequency of the lift pump, Hz; $H$ is the total head, m; $H_1$ is the rated head of the lift pump, m; among them, the calculation formula for the total head is:

[0091] $H = H_2 - H_3 + H_{fa}$;

[0092] In the formula, $H_2$ is the pumping height of the pump house, m; $H_3$ is the liquid level of the pump house, m; $H$ fa is the actual hydraulic loss of the pipeline; among them, the calculation formula for the actual hydraulic loss of the pipeline is:

[0093]

[0094] In the formula, $Q_1$ is the predicted value of the optimal lifting water volume, $m$ 3 / h; $Q_0$ is the rated flow of the lift pump, $m$ 3 / h; $H$ f is the designed hydraulic loss, m; among them, the calculation formula for the designed hydraulic loss is:

[0095] $H_f = H_1 - H_2 + H_4$;

[0096] In the formula, $H$ f is the designed hydraulic loss, m; $H_1$ is the rated head of the lift pump, m; $H_2$ is the pumping height of the pump house, m; $H_4$ is the designed liquid level of the pump house, m. Correcting the designed hydraulic loss to obtain the actual hydraulic loss of the pipeline can improve the accuracy of the data.

[0097] Furthermore, in step S5, the calculation formula for the number of operating units is:

[0098]

[0099] In the formula, f a is the optimal operating frequency of the lift pump, Hz; Q1 is the predicted value of the optimal lifting water volume, m 3 / h; Q0 is the rated flow rate of the lift pump, m 3 / h.

[0100] Further, before performing step S3, outliers in each column of the data sample set are removed first.

[0101] Further, the interquartile range (IQR) criterion is used to remove outliers. The specific steps are as follows: Calculate the first quartile and the third quartile for each column of data, and obtain the interquartile range based on the difference between the two; then calculate the upper and lower bounds of the outliers according to the first quartile, the third quartile, and the interquartile range; finally, remove the outliers in each column of data that are less than the lower bound and greater than the upper bound from the data set.

[0102] Other features, working principles, and beneficial effects of this embodiment are the same as those of Embodiment 1.

[0103] Embodiment 3

[0104] This embodiment is the first embodiment of a control system for a lift pump, which is used to execute the control method of Embodiment 2. It includes: a data acquisition module for collecting data; a hydraulic parameter calculation module, signal-connected to the data acquisition module, for receiving the historical data collected by the data acquisition module and calculating the optimal operating frequency; a data optimization module, signal-connected to the data acquisition module and the hydraulic parameter calculation module respectively, for removing outliers from the data; an SSA-SVR prediction module, signal-connected to the data acquisition module, the hydraulic parameter calculation module, and the data optimization module respectively, for receiving data to train the model and predicting the optimal lifting water volume; a decision-making module for receiving the data of the predicted value of the optimal lifting water volume and calculating the corresponding number of lift pumps to be started; an equipment control module, signal-connected to the SSA-SVR prediction module and the decision-making module respectively, for being signal-connected to the lift pump and controlling the number of lift pumps to be started and the operating frequency.

[0105] Other features, working principles, and beneficial effects of this embodiment are the same as those of Embodiment 2.

[0106] In the specific content of the above specific implementation manners, each technical feature can be combined arbitrarily without contradiction in order to make the description concise, not all possible combinations of the above technical features are described. However, as long as the combinations of these technical features do not conflict, they should all be considered as the scope recorded in this specification.

[0107] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limitations on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation manners here. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. A control method for a lift pump, characterized in that, It includes the following steps: S1: Collect historical data of multiple groups of pump house liquid levels, influent water quality parameters, and effluent water quality parameters; S2: Calculate the optimal lifting water volume under the pump house liquid level to obtain a data sample set; S3: Use the pump house liquid level, influent water quality parameters, and effluent water quality parameters in the data sample set as input features, and the optimal lifting water volume as the output feature to train the SSA-SVR prediction model; S4: The SSA-SVR prediction model predicts the corresponding optimal lifting water volume prediction value according to the new pump house liquid level, influent water quality parameters, and effluent water quality parameter data; S5: Calculate the optimal operating frequency of the lift pump corresponding to the new pump house liquid level; then calculate the number of lift pumps to be started according to the optimal operating frequency, the optimal lifting water volume prediction value, and the rated flow of the lift pump; S6: Regulate the lift pump according to the optimal operating frequency and the number of pumps started.

2. The control method of a lift pump according to claim 1, characterized in that In the step S1, the influent water quality parameters and the effluent water quality parameters both include chemical oxygen demand concentration, ammonia nitrogen concentration, and suspended solid concentration.

3. The control method of a lift pump according to claim 1, wherein, In the step S3, the training steps of the SSA-SVR prediction model are: S31: Divide the data sample set into a training set and a test set; S32: Use the pump house liquid level, influent water quality parameters, and effluent water quality parameters in the data sample set as input features, and the optimal lifting water volume as the output feature; Use the Sparrow Search Algorithm (SSA) to initialize the population, calculate the mean square error of each sparrow individual through the SVR model; update the sparrow population position according to the behaviors of the discoverer, follower, and scouter of the sparrow; then repeat the position update and mean square error calculation until the maximum number of iterations is reached or the mean square error value converges; finally, use GridSearchCV that defines the hyperparameter range for cross-validation near the approximate optimal solution found by SSA to find the global optimal solution, and use the global optimal solution as the optimal hyperparameters of the SVR model; S33: Retrain the SVR model using the optimal hyperparameters; S34: Calculate the mean square error, mean absolute error, and determination coefficient of the model based on the test set to evaluate the performance of the model. If the model performance meets the requirements, select it as the final SSA-SVR prediction model; otherwise, adjust the model parameters and return to step S31 again.

4. The control method of a lift pump according to claim 3, characterized in that, In the step S32, the SVR model uses the mean square error function to measure the fitness, and the specific algorithm is: Where yi is the true value; is the predicted value.

5. The control method of a lift pump according to claim 4, characterized in that, In the step S32, the SSA algorithm imitates the habits of sparrows, and uses three formulas to define the discoverer, follower, and scouter respectively, combining the effects of global search, local search, and regularization optimization to optimize the hyperparameters in the SVR. Among them, the definition formula of the discoverer is: X i = X i + Q * L for other cases; The definition formula of the follower is: X i = X p + |X i - X p | * A + * L Other cases; The definition formula of the scouter is: X i = X best + β|X best - X i |; Where X i is the position of the i-th sparrow in the iteration; X best is the position of the current optimal individual; T is the maximum number of iterations; R is a random number with a value range of [0, 1], used to determine whether to update the position; ST is the safety threshold with a value range of [0.5, 1], used to determine whether to update the position; Q is a random number that follows the standard normal distribution N(0, 1), used to adjust the amplitude of position update; L is the identity matrix, used to adjust the direction of position update; n is the total number of sparrows; X p is the position of the discoverer in the previous iteration; A + is a random matrix with element value range of [-1, 1], used to adjust the direction of position update; X worst is the position of the worst individual; both α and β are step size control parameters, which are random numbers that follow the normal distribution with a mean of 0 and a variance of 1.

6. The control method of a lift pump according to claim 1, characterized in that, In the step S5, the calculation formula of the optimal operating frequency of the lift pump is: In the formula, f is the optimal operating frequency of the lift pump, Hz; H is the total head of the lift pump, m; H1 is the rated head of the lift pump, m; among them, the calculation formula of the total head is: H = H2 - H3 + Hf a ; Wherein, H2 is the pumping station lifting height, in m; H3 is the pumping station liquid level, in m; H fa is the actual pipeline hydraulic loss; wherein, the calculation formula of the actual pipeline hydraulic loss is: Wherein, Q1 is the predicted value of the optimal water lifting volume, m 3 / h; Q0 is the rated flow rate of the lift pump, m 3 / h; H f is the designed hydraulic loss, m; wherein, the calculation formula of the designed hydraulic loss is: Hf = H1 - H2 + H4; where H f is the designed hydraulic loss, m; H1 is the rated head of the lift pump, m; H2 is the lift height of the pump house, m; H4 is the designed liquid level of the pump house, m.

7. A control method for a lift pump according to claim 6, characterized in that, In the step S5, the calculation formula of the number of pumps started is: where f a is the optimal operating frequency of the lift pump, Hz; Q1 is the predicted value of the optimal lift water volume, m 3 / h; Q0 is the rated flow rate of the lift pump, m 3 / h.

8. The control method of a lift pump according to claim 1, characterized in that, Before performing the step S3, outliers in each column of the data sample set are removed first.

9. The control method of a lift pump according to claim 8, characterized in that, The interquartile range (IQR) criterion is used to remove outliers. The specific steps are as follows: Calculate the first quartile and the third quartile for each column of data, and obtain the interquartile range based on the difference between the two; then calculate the upper and lower bounds of the outliers according to the first quartile, the third quartile, and the interquartile range; finally, remove the outliers in each column of data that are less than the lower bound and greater than the upper bound from the data set.

10. A control system for a lift pump, which is used to execute the control method according to any one of claims 1 to 9, characterized in that, Including: A data acquisition module for acquiring data; A hydraulic parameter calculation module, which is signal-connected to the data acquisition module, for receiving the historical data acquired by the data acquisition module and calculating the optimal operating frequency; A data optimization module, which is respectively signal-connected to the data acquisition module and the hydraulic parameter calculation module, for removing outliers in the data; An SSA-SVR prediction module, which is respectively signal-connected to the data acquisition module, the hydraulic parameter calculation module, and the data optimization module, for receiving data to train the model and predicting the optimal water lifting volume; A decision-making module, which is signal-connected to the SSA-SVR prediction module, for receiving the data of the optimal water lifting volume prediction value and calculating the corresponding number of activated lift pumps; An equipment control module, which is respectively signal-connected to the SSA-SVR prediction module and the decision-making module, for being signal-connected to the lift pumps and controlling the number of activated lift pumps and the operating frequency.

Images