An energy-saving operation system for water pump unit equipment based on the Internet of Things

Abstract

The present invention discloses an energy-saving operation system for a water pump unit device based on the Internet of Things, which includes a data acquisition module, an edge computing gateway, a cloud platform, an energy efficiency evaluation module, and a scheduling optimization module connected in sequence. The data acquisition module collects data from the Internet of Things pumping station unit and the intelligent power acquisition monitoring device and uploads it to the cloud platform through the edge computing gateway. The energy efficiency evaluation module uses the collected data on the cloud platform to construct a water pump energy efficiency evaluation model to determine the energy-saving energy efficiency level of the water pump unit operation. When the energy efficiency level is level 3, the scheduling optimization module establishes a model of a constrained optimization problem with the goal of minimizing the daily power consumption cost, and uses an improved differential evolution algorithm to solve the model of the constrained optimization problem, obtaining the speed regulation rate and start-stop operation status quantity of each water pump at each time period, and further controlling the frequency of the frequency converter and the start-stop of the Internet of Things pumping station unit. The present invention improves the energy efficiency potential of the water pump unit device and achieves the economic energy-saving benefit.

Description

Technical Field

[0001] The present invention relates to the field of energy efficiency monitoring for the operation of pump units, and more particularly to an energy-saving operation system for pump unit equipment based on the Internet of Things. Background Art

[0002] As an important energy conversion device and fluid transportation equipment, pumps are widely used general machinery. According to statistics, the total annual power consumption of pump equipment in China reaches 220 billion kWh, accounting for about 20% of the total national power consumption and about 30% of the industrial power consumption. At present, the operating efficiency of pump equipment in China is 10% - 30% lower than that in developed countries. It is not feasible to achieve energy conservation solely by improving the efficiency of pump products, and the energy-saving situation is rather severe.

[0003] At present, most of the pumps in operation in China use valve throttling to adjust the flow rate to meet the constantly changing flow requirements. This method is an uneconomical operation mode. Research shows that using variable speed regulation and controlling the start and stop of pumps are effective methods to avoid throttling losses, and there is great potential for power saving. Summary of the Invention

[0004] In view of this, the present invention provides an energy-saving operation system for pump unit equipment based on the Internet of Things.

[0005] The present invention achieves the above technical objectives through the following technical means.

[0006] An energy-saving operation system for pump unit equipment based on the Internet of Things, comprising:

[0007] A data acquisition module for collecting data from Internet of Things pump stations and intelligent power acquisition and monitoring devices;

[0008] An edge computing gateway for performing edge computing on the collected data and then uploading it to the cloud platform;

[0009] An energy efficiency evaluation module for constructing a pump energy efficiency evaluation model using the collected data on the cloud platform to determine the energy-saving energy efficiency level of the pump unit operation;

[0010] A scheduling optimization module, when the energy efficiency level is 3, establishing a model of a constrained optimization problem with the goal of minimizing the daily power consumption cost, and using an improved differential evolution algorithm to solve the model of the constrained optimization problem to obtain the speed regulation rate and start-stop operation status quantity of each pump for each time period;

[0011] A control output module for obtaining the speed regulation rate and start-stop operation status quantity through the edge computing gateway and controlling the frequency of the frequency converter and the start and stop of the Internet of Things pump station unit.

[0012] In the above technical solution, the model of the constrained optimization problem is specifically:

[0013]

[0014] The constraint terms of the model include:

[0015] w ij ∈ {0, 1}

[0016]

[0017] 0.5 ≤ k ij ≤ 1

[0018]

[0019] Where: C sum is the total electricity cost of the IoT pump unit in a day. There are m parallel - running pumps in the IoT pump station unit, and a day is divided into n equal time periods with a time - step of Δt; ρ is the liquid density, g is the acceleration due to gravity, H dem is the head demand of the unit, w ij is the start - stop operation status variable of the i - th pump in the j - th time period, c j is the unit electricity price in the j - th time period, Q ij is the flow rate of the i - th pump in the j - th time period, k ij is the speed regulation rate of the i - th pump in the j - th time period, η 0 ′ is the efficiency of the pump operating at the rated speed, Q total is the total demand flow rate of the pump station.

[0020] In the above - mentioned technical solution, the improved differential evolution algorithm for solving the model of the constrained optimization problem is specifically as follows:

[0021] ① Population initialization: The number of vector individuals in each generation of the population is m·n, and the vector individual dimension D is 2. Initialize the speed regulation rate and the start - stop operation status variable, and perform a rounding operation on the start - stop operation status variable;

[0022] ② Mutation: For each objective vector individual in the G - th generation of the population, through the mutation operation, generate the corresponding mutant vector individual, and use a time - varying scaling factor to ensure the balance between the diversity of the population and the global optimization ability and convergence speed;

[0023] ③ Crossover: Use the crossover operation to increase the diversity of the interference parameter vectors and generate the trial vector individuals;

[0024] ④ Calculate the fitness

[0025] ⑤ Selection: Compare X i (G) and U i(G)’s fitness value, retain the vector individuals with smaller fitness values in the next-generation population; calculate the fitness value of each vector individual in the contemporary population, record the vector individual with the smallest fitness value, and then compare it with the vector individuals that appear in the next-generation population, retaining the vector individual with the smaller fitness value among the two; where, X i (G) is the i-th target vector individual of the G-th generation population, U i (G) is the i-th test vector individual of the G-th generation population;

[0026] Perform cyclic iteration on ② - ⑤ until the iteration ends. In the final-generation population, the vector individual with the smallest fitness value is the optimization result, that is, the speed regulation rate and the start-stop operation status quantity of each pump at each time period.

[0027] In the above technical solution, the initialization of the speed regulation rate and the start-stop operation status quantity, and the rounding operation on the start-stop operation status quantity are specifically as follows:

[0028]

[0029]

[0030] where, x i,1 is the initial-generation population of the start-stop operation status quantity, are the upper and lower limits of x i,1 respectively, random.uniform(0,1) represents a random real number in the interval [0,1], np.round represents rounding to the nearest integer, x i,2 is the initial-generation population of the pump speed regulation rate, are the upper and lower limits of x i,2 respectively.

[0031] In the above technical solution, the corresponding mutant vector individual is:

[0032] v i，1 = math.ceil(x i，1 + F(x best - x r3 ) + F(x r1 - x r2 ))

[0033] v i,2 = x i,2 + F(x best - x r3 ) + F(x r1 - x r2 )

[0034] F = F 0 × 2 λ ,

[0035] Among them, v i,1 and v i,2 are mutant vector individuals, x best is the optimal vector individual of the current generation, x r1 , x r2 , x r3 are three randomly selected vector individuals, F is a time-varying scaling factor, F 0 is a constant, G is the current iteration number, G t is the maximum iteration number, λ is an intermediate quantity, and math.ceil represents rounding up.

[0036] In the above technical solution, the crossover operation is specifically as follows:

[0037]

[0038] Among them: u i,j is the population individual vector after the crossover operation, that is, the trial vector individual; j rand is an integer serial number randomly selected within the range of [1, D], and CR is the crossover probability; j = 1, 2.

[0039] In the above technical solution, the CR adopts a time-varying crossover method:

[0040]

[0041] Among them, CR max and CR min are the maximum and minimum crossover probabilities respectively.

[0042] In the above technical solution, the fitness satisfies:

[0043]

[0044] Among them: ω, θ, and φ are penalty factors, δ Q , δ k and δ w are constraint quantization results, and

[0045] δ Q = |Q total - Q ij |

[0046]

[0047]

[0048] In the above technical solution, the process of establishing the pump energy efficiency evaluation model is as follows: taking the pump operation flow rate, the specific speed of the pump, and the current pump efficiency as inputs, and the energy efficiency benchmark value and the energy efficiency correction value as outputs, a recurrent neural network is established, and the recurrent neural network is trained using training samples.

[0049] In the above technical solution, the test samples are input into the pump energy efficiency evaluation model, and the corresponding energy efficiency benchmark value and energy efficiency correction value are output. Substituting them into the energy efficiency evaluation grade determination table, three energy efficiency grades are determined, and then the specific energy efficiency grade is determined according to the current pump efficiency; the energy efficiency evaluation grade determination table is divided by the pump operation flow rate, the specific speed of the pump, the energy efficiency benchmark value, and the energy efficiency correction value, in combination with the current pump energy efficiency standards and specifications.

[0050] The beneficial effects of the present invention are as follows: the energy efficiency evaluation module in the present invention uses the collected data on the cloud platform to construct a pump energy efficiency evaluation model to determine the energy-saving energy efficiency grade of the pump unit operation. When the energy efficiency grade is level 3, the scheduling optimization module establishes a model of a constrained optimization problem with the goal of minimizing the daily power consumption cost, and uses an improved differential evolution algorithm to solve the model of the constrained optimization problem, obtaining the speed regulation rate and the start-stop operation state quantity of each pump at each time period, and then controlling the frequency of the frequency converter and the start and stop of the Internet of Things pumping station unit, realizing energy saving of the pump unit equipment, which has important significance and role for improving the energy utilization rate of the pump system. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic structural diagram of the energy-saving operation system of the pump unit based on the Internet of Things according to the present invention;

[0052] Figure 2 It is a flow chart of the differential evolution algorithm according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiments of the present invention will be described completely below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be noted that the embodiments of the present invention are illustrative, but not restrictive of the present invention, so the present invention is not limited to the above embodiments. Based on the principle of the present invention, all other implementation manners obtained by those of ordinary skill in the art without creative efforts are regarded as within the protection scope of the present invention.

[0054] As Figure 1 shown, an energy-saving operation system of a pump unit based on the Internet of Things according to the present invention includes an Internet of Things pumping station unit, an intelligent power collection and monitoring device, an edge computing gateway, a control output module, a data collection module, an energy efficiency evaluation module, a scheduling optimization module, and a cloud platform.

[0055] In this embodiment, the intelligent power acquisition and monitoring device uses the USR-DR154 of the manned Internet of Things, and the data acquisition module uses the NI USB-6343 signal acquisition card.

[0056] The data acquisition module collects data of the Internet of Things pump station units and the intelligent power acquisition and monitoring device by accessing the DI / AI signal source and the Modbus transmitter, including but not limited to the three-phase current of the motor, the three-phase voltage of the motor, the total active power of the motor, the operation flow of the pump, the liquid flow rate inside the pump, the output speed of the motor, the output torque of the motor, and the inlet and outlet pressures of the pump.

[0057] The edge computing gateway is connected to the data acquisition module, performs edge computing on the collected data, and then realizes the rapid upload of the collected data to the cloud platform through 4G network or network port data transparent transmission, and can be monitored in real time on the mobile phone side and the PC side. Specifically, the mobile phone side and the PC side communicate with the cloud platform.

[0058] The control output module completes the control of the start-stop switch and the frequency converter frequency of the Internet of Things pump station unit according to the output of the scheduling optimization module. The start-stop switch is set in the on-site industrial control box and is used to control the start and stop of the motor (pump).

[0059] The cloud platform performs cloud computing through the energy efficiency evaluation module and the scheduling optimization module.

[0060] The energy efficiency evaluation module first calculates the specific speed n of the pump according to the collected data on the cloud platform s :

[0061]

[0062] Secondly, calculate the current pump efficiency η 0 :

[0063]

[0064] Among them, Q is the operation flow of the pump (m 3 / h), ρ is the liquid density (kg / m 3 ), g is the acceleration of gravity (m / s 2 ), T is the output torque of the motor (Nm), n is the speed of the motor (r / min), H is the operation head of the pump (m), and can be obtained by the following formula:

[0065]

[0066] Among them, p 1 , p 2 are the inlet and outlet pressures of the pump (Pa) respectively, z 1 , z 2 are the heights of the inlet and outlet of the pump (m) respectively, v1 and v 2 are the liquid flow velocities (m / s) at the inlet and outlet of the pump respectively.

[0067] Based on the operating flow rate Q of the water pump, the specific speed n of the water pump s , the energy efficiency benchmark value η and the energy efficiency correction value Δη, combined with the current water pump energy efficiency standards and specifications, evaluate the energy-saving degree of the water pump unit operation, and divide it into energy efficiency level 1, energy efficiency level 2, and energy efficiency level 3, corresponding to three evaluation levels of energy-saving, general energy-saving, and non-energy-saving respectively, forming an energy efficiency evaluation level determination table as shown in Table 1;

[0068] Table 1 Energy Efficiency Evaluation Level Determination Table

[0069]

[0070] Take the operating flow rate Q of the water pump, the specific speed n of the water pump s and the current water pump efficiency η 0 as inputs, and the energy efficiency benchmark value η and the energy efficiency correction value Δη as outputs to establish a recurrent neural network (RNN); divide the flow rate, flow velocity, motor output speed, motor output torque, and water pump inlet and outlet pressure on the cloud platform into training samples and test samples. The training samples are used to train the recurrent neural network to obtain a water pump energy efficiency evaluation model; the test samples are input into the water pump energy efficiency evaluation model, and the corresponding energy efficiency benchmark value η and energy efficiency correction value Δη are output, substituted into Table 1 to determine the three energy efficiency levels, and then according to the current water pump efficiency η 0 , determine the specific energy efficiency level.

[0071] The scheduling optimization module optimizes the unit operation scheduling with reference to the energy efficiency evaluation results (i.e., when the energy efficiency level is level 3), and formulates the most economical and energy-saving unit scheduling strategy as: establish a model of a constrained optimization problem (Formula (1)) with the goal of minimizing the daily electricity consumption cost, use the improved differential evolution algorithm to solve the model of the constrained optimization problem, obtain the speed regulation rate and start-stop operation status quantity of each water pump in each time period, and send the control signal to the edge computing gateway through the cloud platform. The edge computing gateway transmits the corresponding digital and analog output quantities to the control output module, and finally completes the control of the Internet of Things pumping station unit.

[0072]

[0073] Among them, C sum is the total electricity cost of the water pump unit equipment in a day. There are m parallel-operating water pumps in the Internet of Things pumping station unit, and a day is divided into n equal time periods with a time step of Δt; c j is the unit electricity price in the jth time period; w ij is the start-stop operation status quantity of the ith water pump in the jth time period; N ijis the input power of the i-th pump in the j-th time period:

[0074]

[0075] where, H ij is the head of the i-th pump in the j-th time period, and H ij = H dem , H dem is the head demand of the unit; Q ij is the flow rate of the i-th pump in the j-th time period; η (vfd)ij , η (motor)ij , η (pump)ij are the efficiencies of the frequency converter, motor, and pump of the i-th pump in the j-th time period respectively;

[0076] The efficiency fitting formula of the frequency converter is expressed as: (Bernier M A, Bourret B. Pumping energy and variable frequency drives. ASHRAE J, 1999, (12): 37.)

[0077] η (vfd) = 0.5087 + 1.283k - 1.42k 2 + 0.5864k 3

[0078] where, k is the speed regulation rate;

[0079] When the speed regulation rate k ∈ (0.4, 1), the change range of the motor efficiency η (motor) is [92%, 94%], and the change is not obvious. Considering it as a constant, the average value η (motor) = 0.93;

[0080] The efficiency of the pump is expressed as: (Wu P, Lai Z, Wu D Z et al. Optimization research of parallel pump system for improving energy efficiency. Journal of Water Resources Planning and Management, 2014.)

[0081]

[0082] Therefore, formula (1) is transformed into:

[0083]

[0084] where, k ijLet \(k_{ij}\) be the speed regulation rate of the \(i\)-th water pump in the \(j\)-th time period, and \(\eta'\) be the efficiency of the water pump operating at the rated speed, which can be regarded as a constant. 0 ′ is the efficiency of the water pump operating at the rated speed and can be regarded as a constant.

[0085] Add several constraint terms to the above model:

[0086] ① First, there is a constraint on the start-stop operation status of the water pump. The start is 1 and the stop is 0, as shown in formula (2):

[0087] w ij ∈{0, 1} (2)

[0088] ② The sum of the flow rates of each parallel-operating water pump is equal to the total required flow rate of the pumping station, as shown in formula (3):

[0089]

[0090] ③ To ensure that the water pump does not operate in a low-efficiency and high-energy-consuming state, it is necessary to control the range of the water pump speed regulation rate, as shown in formula (4):

[0091] 0.5 ≤ k ij ≤ 1 (4)

[0092] ④ Frequent start-stop will cause significant wear to the pump and motor, and will also have a certain impact on the pipeline and power system. Therefore, the start-stop switching times of the pump should be minimized, which is characterized by the start-stop operation status, as shown in formula (5):

[0093]

[0094] Use the penalty function method to quantify the constraints in formulas (3)-(5), as shown in formulas (6)-(8):

[0095] δ Q =|Q total -Q ij | (6)

[0096]

[0097]

[0098] Introduce ω, θ, and φ as penalty factors to weigh the influence of the constraints on the optimization problem. The overall constraint penalty function can be expressed as formula (9):

[0099] P(δ) = ωδ Q + θδ k + φδ w (9)

[0100] The traditional differential evolution algorithm (DE) still has deficiencies in solving optimization problems with integer variables. Therefore, DE is improved, and the specific operation steps are as follows:

[0101] ① Population initialization: The number of vector individuals NP in each generation of the population is m·n, and the vector individual dimension D is 2. In the model of the above constrained optimization problem, the integer variable is the start-stop operation state quantity of the water pump. It is necessary to perform rounding operations on the variables based on the original initialization. The initialization pseudo-code formula is as follows:

[0102]

[0103]

[0104] where x i,1 is the initial generation population of the start-stop operation state quantity, are the upper and lower limits of x i,1 respectively. random.uniform(0,1) represents a random real number in the interval [0,1], np.round represents rounding to the nearest integer, and x i,2 is the initial generation population of the water pump speed regulation rate, are the upper and lower limits of x i,2 respectively.

[0105] ② Mutation: For each target vector individual in the Gth generation population, the corresponding mutation vector individual can be generated through the following two formulas. The mutation operation adopts the DE / rand to best / 1 strategy, and a time-varying scaling factor is used to ensure the diversity of the population and the balance between global optimization ability and convergence speed:

[0106] v i，1 = math.ceil(x i，1 + F(x best - x r3 ) + F(x r1 - x r2 ))

[0107] v i，2 = x i，2 + F(x best - x r3 ) + F(x r1 - x r2 )

[0108] F = F 0 × 2 λ ,

[0109] where: v i,1 and v i,2 are the mutation vector individuals, and x bestis the optimal vector individual of the current generation, x r1 , x r2 , x r3 are three randomly selected vector individuals, F is a time-varying scaling factor, and the constant F 0 takes 0.5, G is the current iteration number, and G t is the maximum iteration number, and λ is an intermediate quantity.

[0110] During the optimization process, if the pump start-stop status quantity corresponding to a certain vector individual is a decimal, in order to ensure meeting the hydraulic requirements of the pumping station units, the math.ceil function is used to represent rounding up.

[0111] ③ Crossover: To generate trial vector individuals, the crossover operation is used to further increase the diversity of the interference parameter vectors to improve the differential mutation search strategy. The pseudocode formula of the crossover operation is as follows:

[0112]

[0113] where: u i,j is the population individual vector after the crossover operation (i.e., the trial vector individual), j rand is a randomly selected integer serial number within [1, D], ensuring that at least one dimension of the trial vector individual is inherited from the mutant vector. If random.uniform(0, 1) is less than or equal to the crossover probability CR or j = j rand , then inherit the mutant vector individual v i,j , otherwise copy x i,j ; the value range of CR is [0, 1]. To ensure the diversity in the initial stage of population optimization and the convergence speed in the later stage of optimization, the improved differential evolution algorithm adopts a time-varying CR crossover method:

[0114]

[0115] where CR max and CR min are the maximum and minimum crossover probabilities respectively.

[0116] ④ Calculate fitness: The fitness function is an augmented function of the objective function and constraints. Combining formula (1) and formula (9), the expression of the fitness function is:

[0117]

[0118] where: F is the fitness function.

[0119] ⑤ Selection: Compare X i (G) and U iThe fitness value of (G), retain the vector individuals that make the fitness function have a smaller value in the next-generation population; at the same time, calculate the fitness value of each vector individual in the current population, record the vector individual with the smallest fitness value, and then compare it with the vector individuals that appear in the next-generation population, and retain the vector individual with the smaller fitness value among the two; where X i (G) is the i-th target vector individual of the G-th generation population, U i (G) is the i-th test vector individual of the G-th generation population, and:

[0120] X i (G) = (x i,1 (G), x i,2 (G)), i = 1, 2,..., NP

[0121] U i (G) = (u i,1 (G), u i,2 (G)), i = 1, 2,..., NP.

[0122] Perform loop iteration on the above steps ② - ⑤ until the maximum number of iterations G t or the difference in the fitness values of the vector individuals in adjacent generations of the population is less than 1×10 -7 ; in the final population, the vector individual with the smallest fitness value is the optimization result, that is, the speed regulation rate and the start-stop operation state quantity of each water pump in each time period. The speed regulation rate is used to control the frequency of the frequency converter, and the start-stop operation state quantity is used for the opening and closing of the motor (water pump), and is transmitted to the control output module in real time to achieve precise control of the pumping station unit equipment. The basic algorithm flow is as Figure 2 shown.

[0123] The described embodiments are the preferred embodiments of the present invention, but the present invention is not limited to the above embodiments. Without departing from the essence of the present invention, any obvious improvements, substitutions or variations that those skilled in the art can make all fall within the protection scope of the present invention.

Claims

1. An energy-saving operation system for pump units based on the Internet of Things, characterized in that, it includes: A data acquisition module for acquiring data of the Internet of Things pump station units and intelligent power acquisition monitoring devices; An edge computing gateway that performs edge computing on the acquired data and then uploads it to the cloud platform; The energy efficiency evaluation module uses the acquired data on the cloud platform to construct a pump energy efficiency evaluation model to determine the energy-saving energy efficiency level of the pump unit operation; The scheduling optimization module, when the energy efficiency level is level 3, establishes a model of a constrained optimization problem with the goal of minimizing the daily power consumption cost, and uses an improved differential evolution algorithm to solve the model of the constrained optimization problem to obtain the speed regulation rate and start-stop operation status quantity of each pump for each time period; The control output module obtains the speed regulation rate and start-stop operation status quantity through the edge computing gateway, and controls the frequency of the frequency converter and the start and stop of the Internet of Things pump station units; The model of the constrained optimization problem is specifically: The constraint terms of the model include: w ij ∈{0,1} 0.5≤k ij ≤1 Where: C sum is the total electricity cost of the IoT pump unit in a day. There are m parallel - running pumps in the IoT pumping station unit, and a day is divided into n equal time periods with a time - step of Δt; ρ is the liquid density, g is the acceleration due to gravity, H dem is the head demand of the unit, w ij is the start - stop operation status variable of the i - th pump in the j - th time period, c j is the unit electricity price in the j - th time period, Q ij is the flow rate of the i - th pump in the j - th time period, k ij is the speed - regulation rate of the i - th pump in the j - th time period, η 0 ′ is the efficiency of the pump operating at the rated speed, Q total is the total demand flow rate of the pumping station.

2. The energy-saving operation system for pump units according to claim 1, characterized in that, The improved differential evolution algorithm for solving the model of the constrained optimization problem is specifically: ① Population initialization: The number of vector individuals in each generation of the population is m·n, the vector individual dimension D is 2, the speed regulation rate and start-stop operation status quantity are initialized, and the start-stop operation status quantity is rounded; ② Mutation: For each target vector individual in the Gth generation of the population, through mutation operation, a corresponding mutant vector individual is generated, and a time-varying scaling factor is used to ensure the balance of population diversity, global optimization ability and convergence speed; ③ Crossover: Use the crossover operation to increase the diversity of the interference parameter vectors and generate trial vector individuals; ④ Calculate fitness ⑤ Selection: Compare X i (G) and U i (G), retain the vector individuals with smaller fitness values in the next-generation population; calculate the fitness values of each vector individual in the current population, record the vector individual with the smallest fitness value, and then compare it with the vector individuals that appear in the next-generation population, and retain the vector individual with the smaller fitness value among the two; where X i (G) is the i-th target vector individual of the G-th generation population, and U i (G) is the i-th trial vector individual of the G-th generation population; Perform cyclic iteration on ②-⑤ until the iteration ends. In the final generation of the population, the vector individual with the minimum fitness value is the optimization result, that is, the speed regulation rate and start-stop operation status quantity of each pump for each time period.

3. The energy-saving operation system for pump units according to claim 2, characterized in that, The specific operation of initializing the speed regulation rate and start-stop operation status quantity and rounding the start-stop operation status quantity is: where x i,1 is the initial generation population of the start-stop operation status quantity, are the upper and lower limits of x i,1 respectively, random.uniform(0,1) represents a random real number in the interval [0,1], np.round represents rounding to the nearest integer, and x i,2 is the initial generation population of the pump speed regulation rate, are the upper and lower limits of x i,2 respectively.

4. The energy-saving operation system for pump units according to claim 3, characterized in that, The corresponding mutant vector individual is: v i，1 = math.ceil(x i，1 + F(x best - x r3 ) + F(x r1 - x r2 )) v i,2 = x i,2 + F(x best - x r3 ) + F(x r1 - x r2 ) F = F 0 × 2 λ , Among them, v i,1 and v i,2 are mutant vector individuals, x best is the optimal vector individual of the current generation, x r1 , x r2 , x r3 are three randomly selected vector individuals, F is a time-varying scaling factor, F 0 is a constant, G is the current iteration number, G t is the maximum iteration number, λ is an intermediate quantity, and math.ceil represents rounding up.

5. The energy-saving operation system for pump units according to claim 4, characterized in that, The specific crossover operation is: where: u i,j is the population individual vector after the crossover operation, i.e., the test vector individual; j rand is an integer serial number randomly selected within [1, D], and CR is the crossover probability; j = 1, 2.

6. The energy-saving operation system for pump units according to claim 5, characterized in that, The CR adopts a time-varying crossover method: Among them, CR max and CR min are the maximum and minimum crossover probabilities respectively.

7. The energy-saving operation system for pump units according to claim 2, characterized in that, The fitness satisfies: where: ω, θ, and φ are penalty factors, and δ Q , δ k , and δ w are constraint quantization results, and δ Q = |Q total - Q ij | 8. The energy-saving operation system for pump units according to claim 1, characterized in that, The establishment process of the pump energy efficiency evaluation model is: taking the pump operation flow rate, the specific speed of the pump and the current pump efficiency as inputs, and the energy efficiency reference value and energy efficiency correction value as outputs, establishing a recurrent neural network, and training the recurrent neural network using training samples.

9. The energy-saving operation system for pump units according to claim 8, characterized in that, The test sample is input into the energy efficiency evaluation model of the water pump, and the corresponding energy efficiency baseline value and energy efficiency correction value are output. Substitute them into the energy efficiency evaluation grade determination table to determine three energy efficiency grades, and then determine the specific energy efficiency grade according to the current water pump efficiency; the energy efficiency evaluation grade determination table is divided based on the water pump operation flow rate, the specific speed of the water pump, the energy efficiency baseline value and the energy efficiency correction value, in combination with the current water pump energy efficiency standards and specifications.

Images