Pump energy saving optimization method and system for complex pump deployment forms

Abstract

The application provides a water pump energy-saving optimization method and system for a complex water pump deployment form, comprising: a data acquisition step: acquiring and obtaining operation data of a water pump system; a model training step: preprocessing the data, and training a model based on a preprocessing result, thereby obtaining a water pump system model; a parameter optimization step: acquiring real-time operation parameters, and optimizing control parameters of the water pump based on the water pump system model; and a parameter issuing step: issuing the control parameters for execution. The application realizes efficient operation and energy saving of a complex water pump system by acquiring water pump operation data and pipeline data in real time.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of water pump or fluid delivery energy saving control, and in particular, relates to a water pump energy saving optimization method and system for complex water pump deployment forms. The complex water pump, i.e. a water pump system capable of dividing water pumps into group a and group b, and parallel connection of water pumps within the group and series connection of water pumps between the groups. More specifically, it relates to a water pump system energy saving optimization method based on multi-sensor data acquisition, data cleaning, Bayesian modeling and optimization algorithm, suitable for complex deployment scenarios such as parallel connection of multiple pumps, series connection of multiple pumps, combination of series and parallel connection, variable frequency speed regulation, etc., aiming to realize efficient operation and energy saving of the water pump system. BACKGROUND

[0002] In industrial water systems, parallel and series connection of water pumps is a common strategy to cope with variable flow demand and improve system redundancy. However, the coordinated operation of different types of water pumps faces significant challenges, especially in terms of output distribution and control optimization. The specific problems are as follows:

[0003] First, uneven output caused by characteristic differences; Specifically, the rated parameters of different types of water pumps, such as flow rate, head, and power curve, differ significantly. When different types of water pumps are connected in parallel, the flow rate-head curve, i.e. Q-H curve, of each pump does not match, leading to a deviation of flow rate distribution from the theoretical optimal value. For example, a high-head pump may "suppress" a low-head pump, causing its operating point to shift to the inefficient region, and even causing backflow, resulting in energy waste. In series connection configuration, if the head-flow characteristics of water pumps are inconsistent, the front-end pump may be overloaded and the rear-end pump may be subject to cavitation, seriously affecting system stability.

[0004] Second, coordination difficulty under dynamic conditions; Specifically, in actual systems, load demand fluctuates over time, requiring dynamic adjustment of the start-stop combination and frequency of water pumps. Traditional methods mostly use fixed priority or rotation strategies, which cannot respond to changes in working conditions in real time. For example, in a parallel system, when a new water pump is added, the operating point of the original water pump will drift due to changes in pipe network resistance, and if the control is lagging, it may lead to a decrease in overall efficiency.

[0005] Third, model mismatch and high optimization complexity; Specifically, existing control strategies are usually based on simplified models of a single water pump, such as the constant efficiency assumption, ignoring the coupling effects of multiple pumps. For a combination of different types of water pumps, traditional models cannot accurately predict the relationship between system-level flow rate, head, and power. In addition, the optimization problem involves a mixed integer nonlinear programming (MINLP) of discrete variables, i.e. the number of start-stop units, and continuous variables, i.e. frequency, which is difficult to solve and has poor real-time performance.

[0006] The limitations of existing solutions are as follows:

[0007] I. Unified frequency control: forces all water pumps to run at the same frequency, ignoring performance differences, resulting in some water pumps running inefficiently for a long time.

[0008] II. Empirical rule adjustment: relies on manual priority setting or allocation ratio, lacks data-driven optimization, and has poor adaptability.

[0009] III. High-precision model dependence: some solutions use CFD simulation or high-fidelity models, which are costly and difficult to apply online.

[0010] Patent document CN111237181A discloses a water pump system operation characteristic online identification and optimization control method, characterized by: step S100, collecting basic information of the water pump system; step S200, collecting real-time data of the water pump system; step S300, constructing a characteristic model of the water pump system according to the basic information and real-time data and updating it in real time; and step S400, controlling the water pump system according to the characteristic model of the water pump system. This solution lacks data-driven optimization, has poor adaptability, and is costly and difficult to apply online.

[0011] This problem needs to be solved urgently. SUMMARY

[0012] In view of the defects in the prior art, the purpose of the present application is to provide a water pump energy-saving optimization method and system for complex water pump deployment forms.

[0013] According to the water pump energy-saving optimization method for complex water pump deployment forms provided by the present application, the following steps are included:

[0014] Data collection step: collect and obtain the operation data of the water pump system;

[0015] Model training step: preprocess the data and train the model based on the preprocessing results to obtain the water pump system model;

[0016] Parameter optimization step: obtain real-time operation parameters and optimize the control parameters of the water pump based on the water pump system model;

[0017] Parameter issuing step: issue the control parameters for execution.

[0018] Preferably, in the data collection step, the operation data includes frequency, power, flow, water pump inlet pressure, and water pump outlet pressure; and the collection frequency of the operation data is less than or equal to 1 minute per time.

[0019] Preferably, in the model training step, the preprocessing, i.e. excluding abnormal data in the operation data; the abnormal data includes: data with negative power, data with flow greater than the rated value, data with negative flow or data with negative differential pressure between the outlet pressure of the water pump and the inlet pressure of the water pump;

[0020] The water pump system model comprises a water pump power model, a water pump flow model and a water pump differential pressure model.

[0021] Preferably, the mathematical expression of the water pump power model is:

[0022]

[0023] wherein, P represents the water pump power; Pn represents the rated power of the water pump; f0 represents the standard frequency default value; k represents the frequency and power similarity index; f represents the frequency; k and b represent trained parameters; P represents the water pump flow;

[0024] In the model training step, the water pump power model is trained using a Bayesian model, comprising:

[0025] Step A1: setting Pn is 75%~110% of the rated power of the water pump; k is a positive value; b is a real number;

[0026] Step A2: based on the setting of step A1, the power to be trained target is evaluated by a Bayesian algorithm to obtain the trained water pump power model; the power to be trained target comprises: k and b;

[0027] The mathematical expression of the water pump flow model is:

[0028]

[0029] wherein, , , are all trained parameters; P is the differential pressure of the water pump;

[0030] In the model training step, the water pump flow model is trained using a Bayesian model, comprising:

[0031] Step B1: setting k3 is a positive value, b is a real value, and k3 is 1~3; ​

[0032] Step B2: based on the settings of step B1, evaluate the to-be-trained parameters by the Bayesian algorithm to obtain the trained water pump flow model;

[0033] The mathematical expression of the water pump differential pressure model is:

[0034]

[0035] wherein, , , , , b2 are to-be-trained parameters;

[0036] In the model training step, the water pump differential pressure model is trained using the Bayesian model, comprising:

[0037] Step C1: setting , , is a positive value, is 1-3, and b2 is 1-3;

[0038] Step C2: based on the settings of step C1, evaluate the to-be-trained parameters by the Bayesian algorithm to obtain the trained water pump differential pressure model.

[0039] Preferably, in the model training step, the input of the water pump system model includes: the number and frequency of water pump starts;

[0040] The output of the water pump system model includes: water pump differential pressure, water pump power and water pump flow;

[0041] In the parameter issuing step, the number, combination and control frequency of water pump starts, i.e. Hz, are issued to PLC or DDC for execution; the combination means grouping the water pumps of the water pump system, with the water pumps in the group connected in parallel and the water pumps between groups connected in series.

[0042] According to the water pump energy-saving optimization system for complex water pump deployment form provided by the application, comprising:

[0043] Data acquisition module: acquires and obtains the operation data of the water pump system;

[0044] Model training module: preprocesses the data and trains the model based on the preprocessing result to obtain the water pump system model;

[0045] Parameter optimization module: obtains real-time operation parameters and optimizes the control parameters of the water pump based on the water pump system model;

[0046] Parameter issuing module: issues the control parameters for execution.

[0047] Preferably, in the data acquisition module, the operation data includes: frequency, power, flow, water pump inlet pressure and water pump outlet pressure; the acquisition frequency of the operation data is less than or equal to 1 minute each time.

[0048] Preferably, in the model training module, the preprocessing is to exclude abnormal data in the operation data; the abnormal data includes: data with negative power, data with flow greater than rated value, data with negative flow or data with negative pressure difference between water pump outlet pressure and water pump inlet pressure;

[0049] The water pump system model includes: a water pump power model, a water pump flow model and a water pump pressure difference model.

[0050] Preferably, the mathematical expression of the water pump power model is:

[0051]

[0052] wherein, represents water pump power; represents water pump rated power; represents a standard frequency default value; represents a frequency and power similarity index; represents frequency; k and b are positive values; respectively represent a trained parameter and another trained parameter; represents water pump flow;

[0053] In the model training module, the water pump power model is trained using a Bayesian model, including:

[0054] Module A1: setting is 75% to 110% of water pump rated power; is 2 to 4; k is a positive value; b is a real number;

[0055] Module A2: based on the setting of module A1, the power to be trained target is evaluated by a Bayesian algorithm to obtain a trained water pump power model; the power to be trained target includes: k and b;

[0056] The mathematical expression of the water pump flow model is:

[0057]

[0058] wherein, , , are all parameters to be trained; P is water pump pressure difference;

[0059] In the model training module, the water pump flow model is trained using a Bayesian model, comprising:

[0060] Module B1: setting is a positive value, is a real value, and k3 is 1-3;

[0061] Module B2: based on the setting of module B1, the to-be-trained parameters are evaluated by a Bayesian algorithm to obtain the trained water pump flow model.

[0062] The mathematical expression of the water pump pressure difference model is:

[0063]

[0064] wherein, , , , , and b2 are to-be-trained parameters.

[0065] In the model training module, the water pump pressure difference model is trained using a Bayesian model, comprising:

[0066] Module C1: setting , , is a positive value, is 1-3, and b2 is 1-3.

[0067] Module C2: based on the setting of module C1, the to-be-trained parameters are evaluated by a Bayesian algorithm to obtain the trained water pump pressure difference model.

[0068] Preferably, the input of the water pump system model comprises the number and frequency of water pump starts.

[0069] The output of the water pump system model comprises the water pump pressure difference, water pump power and water pump flow.

[0070] In the parameter issuing module, the number, combination and control frequency (Hz) of water pump starts are issued to PLC or DDC for execution; the combination is that the water pumps of the water pump system are grouped, the water pumps in the group are connected in parallel, and the water pumps between groups are connected in series.

[0071] Compared with the prior art, the present application has the following beneficial effects:

[0072] 1. The present application solves the problems of low energy efficiency and optimization lag of complex water pump systems by fusing mechanism models and data-driven methods, has significant energy-saving effect, has strong robustness and real-time response capability, and can be widely applied to central air conditioning, industrial circulating water, chemical fluid transportation and other scenes.

[0073] 2、The water pump system optimization method and system provided by the application can realize efficient operation and energy saving of a complex water pump system by collecting water pump operation data and pipeline data in real time, constructing a water pump system model and optimizing energy saving parameters. BRIEF DESCRIPTION OF DRAWINGS

[0074] Other features, objects and advantages of the application will become more apparent from the following detailed description of non-limiting embodiments with reference to the attached drawings:

[0075] Figure 1 A water pump system optimization process schematic diagram is provided for the application;

[0076] Figure 2 A pressure difference and flow calculation process schematic diagram of the water pump system optimization in a complex series-parallel form is provided for the application. DETAILED DESCRIPTION

[0077] The application will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the application, but do not limit the application in any form. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the application. These all belong to the protection scope of the application.

[0078] The application discloses a water pump energy saving optimization method and system suitable for a complex water pump deployment form, and belongs to the technical field of industrial energy saving control. The water pump operation data and pipeline data are collected in real time, a water pump system model is constructed, and energy saving parameters are optimized.

[0079] Referring to Figure 1 , the method comprises the following steps:

[0080] First, install the necessary sensors for the task, and collect water pump operation data in real time through PLC or DDC, including frequency, power, flow and inlet and outlet pressure, and store them at a frequency of 1 minute;

[0081] Second, use historical data and clean and select based on the moving window method, combine the physical characteristics of the water pump with Bayesian probability inference to construct a hybrid model, calibrate the parameters through maximum posterior estimation, and accurately depict the relationship between water pump power, flow and pressure difference;

[0082] Then, a differential evolution algorithm is used for multivariate optimization, frequency and the number of start-ups are used as inputs, power minimization is used as the target, flow and pressure difference are constrained to be within the upper and lower limit requirements, and the optimal control strategy is solved;

[0083] Finally, the optimization results are issued to the execution unit through the communication protocol to realize closed-loop dynamic control.

[0084] In other words, the application discloses a water pump energy-saving optimization method and system suitable for complex water pump deployment forms, and comprises the following steps.

[0085] A sensor installation and data collection step: necessary sensors are installed in the water pump system, connected to a PLC or DDC controller, and the total data of the water pump system is collected from the PLC or DDC. The data is collected and stored at a time interval of less than or equal to 1 minute.

[0086] A model training step: first, the historical data is preprocessed to eliminate outliers. Then, according to the model form of the hybrid mechanism, the model parameters are estimated using the Bayesian model training method, and then the water pump system model is obtained by combining the water pump monomer model according to the water pump topology. In other words, the model parameters are estimated using the Bayesian model training method to obtain the water pump system model.

[0087] A parameter optimization step: the current actual pressure / differential pressure, flow and other operating parameters are obtained, the maximum / minimum pressure / differential pressure control range is obtained, and the differential evolution algorithm is used to optimize the number of water pump start-ups, combinations and control frequencies. Under the condition of meeting the differential pressure constraint range, the number of water pump start-ups, combinations and control frequencies with the lowest energy consumption are found.

[0088] An optimized parameter issuing step: the number of water pump start-ups, combinations and control frequencies are issued to the PLC or DDC for execution. And the parameter optimization and optimized parameter issuing are executed every 15 minutes.

[0089] Specifically, in the sensor installation and data collection step, it is necessary to determine whether the following sensors exist, and if not, they need to be installed:

[0090] A pressure sensor: used to monitor the pipeline pressure, and the minimum number of installed sensors needs to meet the following conditions:

[0091] 1. If the water pump is only in parallel deployment form, one pressure sensor needs to be installed on the inlet main pipeline and the outlet pipeline of the water pump;

[0092] 2. If the water pump is divided into two groups, the group is in parallel form, and the group and the group are in series form, then one pressure sensor needs to be installed on the inlet main pipeline and the outlet pipeline of the water pump system, and one pressure sensor needs to be installed on the main pipeline between the two groups of water pumps;

[0093] Specifically, the grouping basis is that the deployment form of the water pumps belonging to the same pipeline section is parallel.

[0094] 3. If the water pump is divided into multiple groups, the group is in parallel, and the group and the group are in complex series-parallel form, then one pressure sensor needs to be installed on the inlet main pipeline and the outlet pipeline of each group of water pumps.

[0095] Flow sensor: for monitoring the pipeline flow, the minimum number of installations needs to meet the following conditions:

[0096] If the water pump is only in parallel deployment form, the water pump inlet main pipeline or outlet pipeline needs to install 1 flow sensor;

[0097] If the water pump is divided into two groups, the group is in parallel form, and the group and the group are in series form, the water pump system's inlet main pipeline or outlet pipeline needs to install 1 flow sensor, and if there is a bypass branch between the two groups of water pumps, the branch needs to install 1 flow sensor;

[0098] If the water pump is divided into multiple groups, the group is in parallel, and the group and the group are in complex series-parallel form, then each group of water pump's inlet main pipeline or outlet pipeline needs to install 1 flow sensor.

[0099] Frequency converter: if the water pump is configured with a frequency converter, the frequency control value of the frequency converter needs to be collected. If there is no frequency converter, the preferred option is to install a frequency converter.

[0100] Electricity meter: each water pump needs to be configured with an electricity meter, and the instantaneous power of the water pump needs to be collected.

[0101] In one possible implementation, in the sensor installation and data collection step, the following data needs to be collected, which is indispensable:

[0102] 1, water pump inlet pressure, that is, P_in;

[0103] 2, water pump outlet pressure, that is, P_out;

[0104] 3, water pump operating frequency, if no frequency converter is installed, the default is power frequency, that is, Hz;

[0105] 4, water pump real-time power, that is, E.

[0106] Specifically, in the model training step, data cleaning is performed first. Data cleaning uses the method of excluding abnormal data, and the excluded data conditions include: negative power, flow exceeding rated value or negative value, and negative pressure difference ΔP = P_out - P_in.

[0107] After excluding abnormal data, the sliding window method is used to screen the collected data for steady state, and remove non-steady state data.

[0108] The sliding window method is to smooth the data by sliding window to eliminate the noise caused by short-term fluctuations.

[0109] The typical implementation steps include: parameter setting: according to the data characteristics and requirements, determine the window size w and the sliding step s. Through testing, w = 10, s = 1 is adopted.

[0110] Window initialization: Set the window start position at the first point of the data sequence.

[0111] Window data processing: Apply window function to the data in the current window, such as calculating mean, median, and standard deviation. Tests show that calculating mean is the best. Compare each data point in the window with the mean, and mark the points that deviate from the mean by more than 2% as outliers for deletion.

[0112] Moving window: Move the window to the right by s, that is, 1 data point, repeat the window data processing step until the entire data set is traversed.

[0113] Result processing: Remove all marked outliers, and the resulting data subset is used to train the model.

[0114] Specifically, in the modeling step, the second step is model training:

[0115] 1. Model form:

[0116] First, Model One: Based on the similarity law formula of water pump , E is the power of the water pump, Hz is the operating frequency of the water pump, and the following model form is used to predict the power of the water pump: , where k, b are trained parameters, The physical meaning of E is the rated power of the water pump, The physical meaning of k is the similarity index of frequency and power, is the standard frequency value, default 50, and (k*F+b) is the correction of flow to power. * indicates multiplication.

[0117] Second, Model Two: Based on the similarity law formula of water pump , F is the flow of the water pump, is the positive proportional symbol, Hz is the operating frequency of the water pump, and the following model is used to predict the flow of the water pump: , , , are trained parameters.

[0118] Third, Model Three: Based on the similarity law formula of water pump , P is the pressure difference of the water pump, Hz is the operating frequency of the water pump, and the following model form is used to predict the pressure difference of the water pump: ,

[0119] where k1, b1, k2, b2 are trained parameters.

[0120] 1. Model training method:​​​​

[0121] The model training using the Bayesian model first determines the coefficient prior range determined by physical knowledge:

[0122] Specifically, in model one, βE_rated should be 75%~110% of the water pump nameplate rated power value, should be between 2~4, k should be a positive value, b should be a real number, and all conform to normal distribution;

[0123] Specifically, in model two, should be a positive value, should be a real value, k3 should be between 1~3, and all conform to normal distribution;

[0124] Specifically, in model three, a1, a2, a3 should all be positive values, b1 should be between 1~3, b2 should be between 1~3, and all conform to normal distribution;

[0125] Then, using the Bayesian method, combining the prior range of the aforementioned parameters and the historical data of no less than 14 days after cleaning, the maximum likelihood estimation of the trained parameters is performed, thereby obtaining all the models of the water pump.

[0126] Specifically, in the modeling step, the third step constructs the system model, the system model input is the number of water pumps turned on, i.e. State and the control frequency, i.e. Hz, and the model output is the differential pressure, i.e. P, the flow, i.e. F and the power, i.e. E.

[0127] Among them:

[0128] (1) Differential pressure (P) = Model three (Hz, F)

[0129] (2) Flow (F) = Model two (Hz, P) * State

[0130] (3) Power (E) = Model one (Hz, F) * State

[0131] Among them, * represents multiplication.

[0132] Since the input of the model is only Hz and State, Newton's method is also used to solve the differential pressure (P) and flow (F) by using P and F to satisfy the above formulas (1) and (2) at the same time. For a system where water pumps are only deployed in parallel, the aforementioned method can adapt to the entire system, and for the other two forms, the following conditions also need to be met:

[0133] When the water pumps are divided into two groups, group a and group b, the groups are in parallel form, and the groups are in series form, additional constraints are required in the calculation process that the group a flow Fa is equal to the group b flow Fb, and the total differential pressure P is the sum of the group a differential pressure Pa and the group b differential pressure Pb.

[0134] When the water pump is divided into multiple groups, the groups are in parallel, and the groups are in complex series-parallel form. The system should comply with the water flow mass conservation, and the pressure difference is gradually increased. The water flow mass conservation is that when the total flow of the system is divided into several series systems, the flows of these series systems are equal, and the sum of the flows of the water pumps in the parallel water pump system in the series system is the total flow of the parallel water pump system. The pressure difference is gradually increased, that is, when the total pressure difference of the system is divided into several series systems, the sum of the pressure differences of these series systems is the total pressure difference of the system. The parallel water pump system in the series system is shown in detail in Figure 2 .

[0135] The above completes the construction of the system model.

[0136] Specifically, in one possible implementation, in the optimization parameter step, the actual water pump opening state, control frequency, pressure difference, flow and other parameters in the current system are obtained, and the upper and lower limits of the pressure difference, the upper and lower limits of the flow and other threshold values are obtained, and the differential evolution algorithm is used for parameter optimization, and the objective function is , that is, the total power consumption of all water pumps is the lowest, and at the same time, P≥pressure difference lower limit, P≤pressure difference upper limit, F≥flow lower limit, and F≤flow upper limit. represents the real-time power of the water pump.

[0137] Differential evolution algorithm, namely Differential Evolution, is a powerful population-based stochastic optimization algorithm, which is particularly suitable for solving global optimization problems in continuous space. It is simple, robust, and has good parallelism, and performs well in dealing with nonlinear, non-differentiable, multi-peak and noisy complex optimization problems. The following are the specific steps of the differential evolution algorithm:

[0138] Step 1: Initialize the population.

[0139] Set parameters: NP is the population size, that is, the number of individuals. Set to 5 to 10 times the problem dimension D, but can be adjusted according to the problem. For the water pump optimization problem, the population size usually needs to be set to 500 or more.

[0140] F is the scaling factor, that is, Scaling Factor. Control the strength of the differential vector disturbance, the range is usually [0.4, 1.0]. F value is large, the exploration ability is strong; F value is small, the development ability is strong. For the water pump optimization problem, the F value is set to 0.4-0.5 according to the test.

[0141] CR is the crossover probability, i.e. Crossover Rate. Control how many components the trial vector inherits from the mutation vector, range in [0, 1]. CR value large, trial vector more like mutation vector; CR value small, trial vector more like target vector. For water pump optimization problem, tested CR value setting in 0.4~0.5 is appropriate.

[0142] G_max is the maximum evolution generation. For water pump optimization problem, tested G_max value setting above 200 is appropriate.

[0143] Randomly generate initial population: randomly generate each individual, for each individual i in the population, i = 1, 2,..., NP: randomly generate a value for each dimension j, j = 1, 2,..., D. Here the dimension is specific: the state of each water pump, i.e. state = 0 or 1, 0 means off, 1 means on; the control frequency of each water pump, the control frequency range is usually [25, 50]. Record this individual as , and set G = 0.

[0144] Calculate the initial population fitness, bring the water pump control parameter information carried by each individual into the system model to calculate F, P, E, i.e. F, P, , f is the system model, then the penalty term correction is made to the calculated E, i.e. , is the penalty function, which takes effect when F exceeds the flow upper and lower limits or P exceeds the pressure difference upper and lower limits, the penalty amount is 50 times the exceeding value, i.e. , ΔF and ΔP are the absolute values of the difference between F and P and the upper and lower limits after exceeding the upper and lower limits.

[0145] Step two: main loop, i.e. G = 0 to G = G_max-1.

[0146] For each individual in the population , called the target vector Target Vector.

[0147] Mutation: generate a mutation vector for the target vector .

[0148]

[0149]

[0150] is the difference vector, which reflects the difference direction and size of two individuals in the solution space. F scales this difference, controls the step size of perturbation. as the base vector of perturbation. If the mutation vector is not in the feasible region, then the mutation vector is replaced by the target vector. some component of goes beyond the bounds , its value is limited to the bound value. is the lowest bound value for the mutation vector, is the highest bound value for the mutation vector. denotes multiplication.

[0151] Crossover: using the mutation vector and the target vector to generate a trial vector . The goal is to introduce population diversity and preserve some information from the target vector. Binary crossover, for each component j of the trial vector, j = 1, 2,..., D: generate a random number . Generate a random integer ; specifically, ensure that at least one component comes from the mutation vector. D is the maximum value of the components.

[0152] For each component j: if or ; specifically, the trial vector takes that component from the mutation vector; otherwise: ; specifically, the trial vector takes that component from the target vector.

[0153] Selection: greedy selection between the trial vector and the target vector decides which one goes to the next generation of the population . If . Specifically, the trial vector is better, and goes to the next generation. Otherwise, ; specifically, the target vector is preserved to the next generation.

[0154] Step three: termination condition check.

[0155] Check and determine if the maximum number of evolutionary generations is reached, and the result is yes, i.e., the termination condition is met, then the algorithm ends, and outputs the individual with the best fitness in the current population as the found approximate global optimal solution. Otherwise, set G = G + 1, and return to the main loop to continue evolution.

[0156] The application also provides a water pump energy-saving optimization system for complex water pump deployment forms, which can be realized by executing the flow steps of the water pump energy-saving optimization method for complex water pump deployment forms, i.e., the water pump energy-saving optimization method for complex water pump deployment forms can be understood by those skilled in the art as the preferred embodiment of the water pump energy-saving optimization system for complex water pump deployment forms.

[0157] According to the application, a water pump energy-saving optimization system for complex water pump deployment forms is provided, comprising:

[0158] A data acquisition module acquires and obtains operation data of the water pump system;

[0159] A model training module preprocesses the data and trains a model based on the preprocessing result, thereby obtaining a water pump system model;

[0160] A parameter optimization module acquires real-time operation parameters and optimizes control parameters of the water pump based on the water pump system model;

[0161] A parameter issuing module issues the control parameters for execution.

[0162] Those skilled in the art know that, in addition to implementing the system and each device, module and unit thereof provided by the application in a pure computer readable program code manner, the system and each device, module and unit thereof provided by the application can also be implemented in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded microcontrollers, etc. by logically programming the method steps to achieve the same functions. Therefore, the system and each device, module and unit thereof provided by the application can be considered as a hardware component, and the devices, modules and units included therein for achieving various functions can also be considered as structures within the hardware component; the devices, modules and units for achieving various functions can also be considered as both software modules for implementing methods and structures within hardware components.

[0163] In the description of the present application, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0164] The specific embodiments of the application have been described above. It should be understood that the application is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essential content of the application. The embodiments of the present application and the features in the embodiments can be arbitrarily combined with each other without conflict.

Claims

1. A method for energy optimization of water pumps for complex water pump deployment forms, characterized by, The method comprises the following steps: a data collection step: collecting and obtaining operation data of the water pump system; a model training step: preprocessing the data and training a model based on the preprocessing result, thereby obtaining a water pump system model; a parameter optimization step: obtaining real-time operation parameters and optimizing control parameters of the water pump based on the water pump system model; a parameter issuing step: issuing the control parameters for execution; in the data collection step, the operation data comprises frequency, power, flow, water pump inlet pressure and water pump outlet pressure; the collection frequency of the operation data is less than or equal to 1 minute each time; the water pump system model comprises a water pump power model, a water pump flow model and a water pump pressure difference model; the mathematical expression of the water pump power model is: wherein, represents the water pump real-time power; represents the water pump rated power; represents the standard frequency default value; represents the frequency and power similarity index; represents the frequency; k and both represent the trained parameters; represents the water pump flow rate; in the model training step, the water pump power model is trained using a Bayesian model, comprising: Step A1 : Set 75% to 110% of the water pump rated power; 2 to 4; k is a positive value; b is a real number; Step A2: based on the setting of step A1, the power to be trained target is evaluated by a Bayesian algorithm to obtain a trained water pump power model; the power to be trained target includes: k and b; the mathematical expression of the water pump flow model is: wherein, , , are parameters to be trained; P is the water pump differential pressure; in the model training step, the water pump flow model is trained using a Bayesian model, comprising: Step B1: Setting is a positive value, is a real value, and k3 is 1 to 3; step B2: based on the settings of step B1, the trained parameters are evaluated by a Bayesian algorithm to obtain the trained water pump flow model; the mathematical expression of the water pump pressure difference model is: wherein, , , , , b2 are both to-be-trained parameters; in the model training step, the water pump pressure difference model is trained using a Bayesian model, comprising: Step C1: Setting , , is a positive value, is 1 to 3, and b2 is 1 to 3; step C2: based on the settings of step C1, the trained parameters are evaluated by a Bayesian algorithm to obtain the trained water pump pressure difference model.

2. A method for energy optimization of water pumps for complex water pump deployment forms according to claim 1, characterized in that, in the model training step, the preprocessing is to exclude abnormal data in the operation data; the abnormal data comprises data with negative power, data with flow greater than the rated value, data with negative flow or data with negative pressure difference between the water pump outlet pressure and the water pump inlet pressure.

3. A method for energy optimization of water pumps for complex water pump deployment forms according to claim 1, characterized in that, in the model training step, the input of the water pump system model comprises the number of water pumps started and the frequency; the output of the water pump system model comprises the water pump pressure difference, the water pump power and the water pump flow; in the parameter issuing step, the number of water pumps started, the combination and the control frequency are issued to a PLC or a DDC actuator for execution; the combination means that the water pumps of the water pump system are grouped, the water pumps in the group are connected in parallel, and the water pumps between the groups are connected in series.

4. A water pump energy optimization system for complex water pump deployment forms, characterized by, The method comprises the following steps: a data collection module: collecting and obtaining operation data of the water pump system; a model training module: preprocessing the data and training a model based on the preprocessing result, thereby obtaining a water pump system model; a parameter optimization module: obtaining real-time operation parameters and optimizing control parameters of the water pump based on the water pump system model; a parameter issuing module: issuing the control parameters for execution; in the data collection module, the operation data comprises frequency, power, flow, water pump inlet pressure and water pump outlet pressure; the collection frequency of the operation data is less than or equal to 1 minute each time; the water pump system model comprises a water pump power model, a water pump flow model and a water pump pressure difference model; the mathematical expression of the water pump power model is: wherein, represents the water pump power; represents the water pump rated power; represents the standard frequency default value; represents the frequency and power similarity index; represents the frequency; k and respectively represent a trained parameter and another trained parameter; represents the water pump flow rate; in the model training module, the water pump power model is trained using a Bayesian model, comprising: Module A1 : setting 75% to 110% of the rated power of the water pump; 2 to 4; k is a positive value; b is a real number; Module A2: based on the setting of module A1, the power to be trained target is evaluated by a Bayesian algorithm to obtain a trained water pump power model; the power to be trained target includes: k and b; the mathematical expression of the water pump flow model is: wherein, , , are parameters to be trained; P is the water pump differential pressure; in the model training module, the water pump flow model is trained using a Bayesian model, comprising: Module B1: setting is a positive value, is a real value, k3 is 1-3; Module B2: based on the setting of module B1, evaluate the to-be-trained parameters by Bayesian algorithm, get the trained water pump flow model; The mathematical expression of the water pump differential pressure model is: wherein, , , , , b2 are both to-be-trained parameters; In the model training module, the water pump differential pressure model is trained using a Bayesian model, including: Module C1 : setting , , is a positive value, is 1 to 3, and b2 is 1 to 3; Module C2: based on the setting of module C1, evaluate the to-be-trained parameters by Bayesian algorithm, get the trained water pump differential pressure model.

5. A water pump energy saving optimization system for complex water pump deployment forms as claimed in claim 4 wherein, In the model training module, the preprocessing, i.e. excluding abnormal data in the operation data; the abnormal data includes: data with negative power, data with flow greater than the rated value, data with negative flow or data with negative differential pressure between the outlet pressure of the water pump and the inlet pressure of the water pump.

6. A water pump energy saving optimization system for complex water pump deployment forms as claimed in claim 4, wherein, In the model training module, the input of the water pump system model includes: the number and frequency of water pump start; The output of the water pump system model includes: water pump differential pressure, water pump power and water pump flow; In the parameter issuing module, the number of water pump starts, combination and control frequency are issued to PLC or DDC actuator for execution; the combination means grouping the water pumps of the water pump system, with the water pumps in the group connected in parallel and the water pumps between groups connected in series.

Images