A method and system for fitting pump characteristic curves based on multi-source data fusion
By constructing a multi-source data fusion model and combining the variable frequency pump characteristic curve expression and flow difference relationship, the limitations of the application scenario and the instability of the calculation results of the pump characteristic curve fitting under normal water supply conditions are solved, and accurate calculation is achieved when the water supply conditions change.
Patent Information
- Application Number
- CN202410096031.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-23
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2044-01-23
AI Technical Summary
Existing pump characteristic curve fitting methods have limited application scenarios under normal water supply conditions, especially for variable frequency pumps, where the fitting results are not accurate enough and not stable or reliable enough.
Based on multi-source pump station data, a multi-source data fusion model is constructed. Combining the pump characteristic curve expression under variable frequency regulation, the relationship between flow rate difference and speed ratio, and the solution is obtained through optimization theory to obtain the optimal fitting result.
Fitting pump characteristic curves under normal water supply conditions does not affect the normal operation of the pumping station and can provide accurate calculation results when water supply conditions change.
Smart Images

Figure CN117932537B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of urban water supply and drainage technology, specifically to a method and system for fitting pump characteristic curves based on multi-source data fusion. Background Technology
[0002] With the acceleration of urbanization in my country, the demand for urban water resources is also increasing. Urban water supply systems include raw water intake systems, water treatment systems, and pipeline distribution systems. The pipeline distribution system is the primary source of energy consumption and carbon emissions in the water supply system, mainly relying on water pumps within the network for power. Optimized scheduling of water supply pumping stations is a key technology for reducing the energy consumption and carbon emissions of pump units.
[0003] Optimized scheduling of water supply pumping stations must comply with operational constraints; otherwise, the resulting scheduling instructions will be unenforceable. This necessitates guidance from the pump's characteristic curves. However, due to the irregularity of the flow channels within the pump, pump manufacturers often only obtain local sample points through a limited number of specialized tests, then plot the full characteristic curve through fitting. Therefore, these pump characteristic curves often only meet a few operating conditions and cannot cover the pump's characteristics at different speeds or frequencies. Furthermore, since pumps are moving equipment, they are susceptible to mechanical wear during operation or undergo significant modifications due to project requirements, rendering existing characteristic curves inadequate to reflect actual conditions. Accurate pump characteristic curves can provide constraints for scheduling optimization and reflect the current state of the pump equipment, guiding equipment upgrades and maintenance.
[0004] Currently, pump characteristic curve fitting is divided into experimental method, model regression method and data-driven method. These methods use different data to predict pump characteristic curves. (1) Experimental method mainly involves shutting down other pumps, measuring the head of a single pump at different flow rates, plotting the flow-head curve, and then using a nonlinear fitting algorithm to determine the pump characteristic curve; (2) Model regression method mainly involves constructing a characteristic curve model to describe the nonlinear relationship between flow rate, head, efficiency and shaft power, and then using pump operation data to estimate parameters to determine the model parameters; (3) Data-driven model uses historical operation data of the pump station to record the combination mode, head and flow rate of parallel pumps in the pump station at different times, and uses intelligent algorithms to determine the optimal characteristic curve parameters of each pump in the pump station.
[0005] However, the experimental method requires shutting down other pumps when determining pump flow rate and head, which is suitable for the pump installation stage but difficult to apply to pumps already in operation. The model regression method requires establishing a theoretical characteristic curve model and then using real data to estimate the model parameters, which places high demands on the established model and the quality of the data used for parameter estimation; otherwise, there will be significant errors. The data-driven method requires a large amount of data for modeling and is suitable for systems with relatively stable operating conditions and equipment conditions. However, the model accuracy will decrease when the collected data contains a lot of noise or when the operating conditions change.
[0006] In view of the above, this application is hereby submitted. Summary of the Invention
[0007] The technical problem this invention aims to solve is the limitation of existing technologies in fitting pump characteristic curves, which cannot be performed under normal water supply conditions, especially for variable frequency pumps; furthermore, the calculation results are not accurate enough and are not stable and reliable. The purpose of this invention is to provide a method and system for fitting pump characteristic curves based on multi-source data fusion. Based on multi-source pump station information data, combined with the expression of pump characteristic curves under variable frequency regulation, the relationship between flow difference and the pump head H and speed ratio S, a multi-source data fusion model is constructed; and the model is solved to obtain the optimal fitting result. This invention can be performed under normal water supply conditions without affecting the normal operation of the pump station; even when water supply conditions change, this invention can still provide accurate calculation results.
[0008] This invention is achieved through the following technical solution:
[0009] In a first aspect, the present invention provides a method for fitting pump characteristic curves based on multi-source data fusion, the method comprising:
[0010] Acquire multi-source pump station data and preprocess the multi-source pump station data to obtain preprocessed multi-source pump station data;
[0011] Based on the preprocessed multi-source pump station data, a first model and a second model are constructed. The first model is a pump characteristic curve fitting model based on the similarity principle, and the second model is the relationship between the flow difference and the head and speed ratio of the variable frequency pump.
[0012] Based on the first and second models, optimization theory is used to fuse them to form a multi-source data fusion model.
[0013] The optimal parameters of the multi-source data fusion model are solved to obtain the optimal fitting result.
[0014] Furthermore, the data for multi-source pumping stations includes static pump data, pre-delivery test data, and historical operating data;
[0015] The static data of the water pump includes the water pump model, water pump type, water pump rated power, and water pump rated speed, etc.
[0016] The pre-delivery test data refers to the pump station characteristic curve under standard operating conditions;
[0017] Historical operating data includes the current status of the water pump and other key operating parameters. The current status of the water pump includes its start-up and shutdown status, and any modifications made to it. Other key operating parameters include the water pump's outlet pressure, flow rate, and energy consumption.
[0018] Furthermore, based on the preprocessed multi-source pumping station data, a first model is constructed, including:
[0019] Based on the characteristic curve equation of a fixed-frequency pump, the expression of the characteristic curve equation for the pump's fixed-frequency operation is determined.
[0020] Based on the proportional law of similarity theory, the characteristic curve equation of the water pump under constant frequency operation is rewritten to obtain the characteristic curve expression of the water pump under variable frequency regulation.
[0021] Based on the experimental data before delivery, the characteristic curve expression of the water pump under variable frequency regulation is fitted to form the first model; based on the first model, the first predicted flow rate is calculated.
[0022] Furthermore, the formula for calculating the first predicted flow (i.e., the first model) is as follows:
[0023]
[0024] Where Q(1) is the first predicted flow rate of the variable frequency pump; H and S are the head and speed ratio of the variable frequency pump, respectively; Hx and Sx are the fitting coefficients of the characteristic curve equation of the variable frequency pump, representing the maximum head of the device and the pipeline resistance coefficient, respectively.
[0025] Furthermore, based on the preprocessed multi-source pumping station data, a second model is constructed, including:
[0026] Based on the first model, calculate the first predicted flow rate of the variable frequency pump, and the flow rate difference between the actual flow rate of the variable frequency pump and the first predicted flow rate;
[0027] Based on the preprocessed historical operating data of the pumping station, a second model is formed by establishing the relationship between the flow difference and the head and speed ratio of the variable frequency pump based on the Gaussian process model; the second predicted flow is calculated based on the second model.
[0028] Furthermore, the formula for calculating the second predicted flow (i.e., the second model) is as follows:
[0029] Q(2)=GPR(H,S,θ)
[0030] Where Q(2) is the second predicted flow rate of the variable frequency pump, which is the difference between the actual flow rate of the variable frequency pump and the first predicted flow rate; GPR is the Gaussian prediction model; H and S are the head and speed ratio of the variable frequency pump, respectively; θ is the Gaussian prediction model parameter.
[0031] Furthermore, the expression for the multi-source data fusion model is:
[0032]
[0033] Where Q is the final predicted flow rate of the variable frequency pump.
[0034] Furthermore, the optimal parameters of the multi-source data fusion model are solved, specifically by using the final predicted flow rate of the variable frequency pump as a constraint to solve the optimal parameters of the objective function of the multi-source data fusion model.
[0035] The objective function of the multi-source data fusion model is:
[0036]
[0037] The constraints are:
[0038]
[0039] Among them, Q T Q represents the total flow rate of the pumping station. i To calculate the predicted flow rate of the i-th pump using the first and second models, θ i The parameters are the Gaussian process model (GPR) parameters for the i-th pump.
[0040] Secondly, the present invention provides a pump characteristic curve fitting system based on multi-source data fusion, which uses the aforementioned pump characteristic curve fitting method based on multi-source data fusion; the system includes:
[0041] The acquisition unit is used to acquire multi-source pumping station data and preprocess the multi-source pumping station data to obtain preprocessed multi-source pumping station data.
[0042] The model building unit includes a first model building subunit and a second model building subunit. The first model building subunit is a pump characteristic curve fitting model based on the similarity principle. It is constructed based on the factory test data of the pump station. The second model building subunit is constructed based on the pre-processed historical data of the pump station. The second model is the relationship between the actual flow difference and the head and speed ratio of the variable frequency pump.
[0043] The fusion model forming unit is used to fuse the first model and the second model to form a multi-source data fusion model; and the final prediction result is obtained based on the multi-source data fusion model.
[0044] The solver unit is used to optimize the parameters of the multi-source data fusion model. By conserving the total flow, the parameters of the second model are adjusted to obtain the optimal fitting result.
[0045] Furthermore, the execution process of the first model construction subunit is as follows:
[0046] Based on the characteristic curve equation of a fixed-frequency pump, the expression of the characteristic curve equation for the pump's fixed-frequency operation is determined.
[0047] Based on the proportional law of similarity theory, the characteristic curve equation of the water pump under constant frequency operation is rewritten to obtain the characteristic curve expression of the water pump under variable frequency regulation.
[0048] Based on the experimental data before delivery, the characteristic curve expression of the water pump under variable frequency regulation is fitted to form the first model; based on the first model, the first predicted flow rate is calculated.
[0049] Furthermore, the execution process of the second model construction subunit is as follows:
[0050] Based on the first model, calculate the first predicted flow rate of the variable frequency pump, and the flow rate difference between the actual flow rate of the variable frequency pump and the first predicted flow rate;
[0051] The relationship between the flow difference and the head and speed ratio of the variable frequency pump is established based on the Gaussian process model, forming a second model; the second predicted flow rate is calculated based on the second model.
[0052] Furthermore, the execution process of fusing the model forming unit and the solution unit is as follows:
[0053] The calculation results of the first model and the second model are added together to form a multi-source data fusion model (i.e., a traffic prediction fusion model);
[0054] Based on actual historical operating data, the parameters of the second model are adjusted to minimize the total flow prediction error, and the optimal parameters are obtained.
[0055] Compared with the prior art, the present invention has the following advantages and beneficial effects:
[0056] This invention discloses a method and system for fitting pump characteristic curves based on multi-source data fusion. Based on multi-source pump station information data, and combining the pump characteristic curve expression under variable frequency regulation, the relationship between flow rate difference and the pump head H and speed ratio S, a multi-source data fusion model is constructed. The model is then solved to obtain the optimal fitting result. This invention can be performed under normal water supply conditions without affecting the normal operation of the pump station; even when water supply conditions change, this invention can still provide accurate calculation results. Attached Figure Description
[0057] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:
[0058] Figure 1 This is a flowchart of a method for fitting water pump characteristic curves based on multi-source data fusion according to the present invention;
[0059] Figure 2 This is a comparative diagram of the results of the time flow rate and multi-source data fusion model (step C method) and the curve fitting model (step B1 method) of the present invention;
[0060] Figure 3 This is a block diagram of a water pump characteristic curve fitting system based on multi-source data fusion according to the present invention. Detailed Implementation
[0061] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.
[0062] The existing technology has limitations in the application scenarios of fitting water pump characteristic curves, and cannot be performed under normal water supply conditions, especially for variable frequency pumps; in addition, the calculation results are not accurate enough and cannot be stable and reliable.
[0063] To address the above problems, this invention designs a method and system for fitting pump characteristic curves based on multi-source data fusion, aiming to overcome the shortcomings of existing technologies in fitting pump characteristic curves, especially for variable frequency pumps. This invention is based on multi-source pump station information data, which mainly comes from three sources: pump static parameters, pre-delivery experimental data, and historical operating data. Combining the pump characteristic curve expression under variable frequency regulation, the relationship between flow rate difference and the pump head H and speed ratio S, a multi-source data fusion model is constructed. The model is then solved to obtain the optimal fitting result. The multi-source data fusion model integrates various information sources into a single model, assigning different weights and correction terms to obtain a suitable fitting result, thus avoiding the influence of a single information source on the model's accuracy.
[0064] This invention can be carried out under normal water supply conditions without affecting the normal operation of the pumping station; even when the water supply conditions change, this invention can still provide accurate calculation results.
[0065] Example 1
[0066] like Figure 1 As shown, this invention provides a method for fitting pump characteristic curves based on multi-source data fusion. The method includes:
[0067] Step A: Obtain multi-source pumping station data and preprocess the multi-source pumping station data to obtain preprocessed multi-source pumping station data;
[0068] Step A specifically includes:
[0069] Step A1: Obtain multi-source pump station data, which includes static pump data, pre-delivery test data, and historical operating data; specifically:
[0070] The static data of the water pump includes the water pump model, water pump type, water pump rated power, and water pump rated speed, etc.
[0071] The pre-delivery test data refers to the pump station characteristic curve under standard operating conditions;
[0072] Historical operating data includes the current status of the water pump and other key operating parameters. The current status of the water pump includes its start-up and shutdown status, and any modifications made to it. Other key operating parameters include the water pump's outlet pressure, flow rate, and energy consumption.
[0073] Step A2: For the static data of the water pump, select the important data (water pump model, water pump type, water pump rated power and water pump rated speed, etc.) and organize the structured data.
[0074] Step A3: For the pre-shipment test data, it is necessary to read the pump's factory characteristic curve to obtain the coordinate information of the flow-head, flow-power, and flow-efficiency points in the graph, and then organize them into structured data.
[0075] Step A4 involves data cleaning of historical data, including subset selection, column renaming, duplicate value removal, missing value imputation, data consistency processing, outlier handling, and time series processing. Missing value imputation primarily uses the linear value method. Assuming we know the coordinates (x0, y0) and (x1, y1), to obtain the value of x at a certain position within the interval [x0, x1] on the straight line, the relevant calculation formula is as follows:
[0076]
[0077] Outlier handling primarily uses the Z-score method to determine how many standard deviations the value is from the mean. The relevant calculation formula is as follows:
[0078]
[0079] Time series processing mainly involves resampling the running data, including upsampling and downsampling. Upsampling increases the frequency of the data, converting it into smaller time intervals, while downsampling decreases the frequency of the data, converting it into larger time intervals.
[0080] Step B: Based on the preprocessed multi-source pump station data, construct the first model and the second model; the first model is a pump characteristic curve fitting model based on the similarity principle, and the second model is the relationship between the flow difference and the head and speed ratio of the variable frequency pump.
[0081] Step B specifically includes:
[0082] Step B1: Construct the first model, which is a water pump characteristic curve fitting model based on the similarity principle;
[0083] The model is fitted using common expressions for pump characteristic curves. The speed regulation method states that within a certain speed range, the pump performance follows a proportionality law, with the ratios of flow rate, head, and shaft power before and after speed change being the first, second, and third powers of the speed ratio, respectively. Based on similarity theory, the pump performance follows a proportionality law:
[0084]
[0085] In the formula, S is the speed ratio of the water pump, and Q e H e N e Q, H, and N represent the flow rate, head, and power of the pump at its rated speed, respectively, while Q, H, and N represent the flow rate, head, and power of the pump at its variable frequency speed, respectively.
[0086] Based on similarity theory, the expression for the characteristic curve of the water pump under variable frequency regulation is obtained;
[0087]
[0088] Based on the pre-delivery experimental data (i.e., the pump's factory characteristic curve data), the pump characteristic curve expression under frequency conversion regulation is fitted to obtain parameters Hx, Sx, a0, a1, a2, b0, b1, and b2, forming the first model. Based on the first model, when the pump head H and the pump speed ratio S are known, the flow rate Q, i.e., the first predicted flow rate, can be calculated using the pump characteristic curve expression under frequency conversion regulation. The formula for calculating the first predicted flow rate (i.e., the first model) is:
[0089]
[0090] Where Q(1) is the first predicted flow rate of the variable frequency pump; H and S are the head and speed ratio of the variable frequency pump, respectively; Hx and Sx are the fitting coefficients of the characteristic curve equation of the variable frequency pump, representing the maximum head of the device and the pipeline resistance coefficient, respectively.
[0091] Step B2: Construct the second model, which is the relationship between the difference between the actual flow rate of the variable frequency pump and the first predicted flow rate and the pump head and speed ratio.
[0092] Step B2 specifically includes:
[0093] Based on historical operating data, the first predicted flow rate of the variable frequency pump is calculated based on the first model, and the flow difference between the actual flow rate and the first predicted flow rate of a specific variable frequency pump is obtained.
[0094] Based on the Gaussian process model, the relationship between the flow difference and the head and speed ratio of the variable frequency pump is established, forming the second model; the expression of the second model (i.e., the relationship between the flow difference and the head and speed ratio of the variable frequency pump) is:
[0095] Q(2)=GPR(H,S,θ)
[0096] Where Q(2) is the second predicted flow rate of the variable frequency pump, which is the difference between the actual flow rate and the first predicted flow rate of the variable frequency pump; GPR is the Gaussian prediction model, and θ is the parameter of the Gaussian prediction model. Specifically:
[0097] Assuming we know the head H and speed ratio S of each pump, and taking n points from historical data, let X = {[H1,S1], [H2,S2],...,[H...S2]} n ,S n ]} T , Y={Q(2)1,Q(2)2,...,Q(2) n} T Gaussian regression assumes that all data follow an n-dimensional normal distribution, so the objective function Y is a stochastic process, where each variable is a random variable. Therefore:
[0098]
[0099] In the formula, Let μ represent a Gaussian process, Σ be the mean vector, and Σ be the covariance matrix of n features. The covariance between features in each dimension can be simulated using a Gaussian process kernel; this invention uses a radial basis function (RBF) kernel. Expanding Σ, we get:
[0100]
[0101]
[0102] In the formula, cov i,j cor represents the covariance of feature i and feature j, t i and t j These correspond to the i-th and j-th features, respectively, and σ and l represent the RBF kernel coefficients.
[0103] Having defined the mean and variance, we can write the conditional probability of Y for a specific value given our defined μ and Σ as follows:
[0104]
[0105] Choose values for μ and Σ to maximize the above equation, take the logarithm of both sides of the equation, and calculate the partial derivatives of μ and Σ respectively. The solution can be obtained by making the derivative zero.
[0106] Step C: Based on the first model and the second model, the optimization theory is used to fuse them to form a multi-source data fusion model and its optimization solution;
[0107] Step C specifically includes:
[0108] Establish a fusion model for predicting the flow rate of a single pump.
[0109]
[0110] In the formula, i represents the i-th variable frequency pump, Q i Hx i Sx i H i S i GPR i θ i Let represent the flow rate, maximum device head, pipeline resistance coefficient, head, speed ratio, Gaussian process model and its parameters for the i-th variable frequency pump, respectively.
[0111] An optimization model is established with the total flow prediction error of the pumping station as the objective function and the flow prediction fusion model of a single pump as the constraint, as follows:
[0112] The objective function is:
[0113]
[0114] The constraints are:
[0115]
[0116] Among them, Q T Q represents the total flow rate of the pumping station. i To calculate the predicted flow rate of the i-th pump using the first and second models, θ i Let be the parameters of the Gaussian process model for the i-th pump.
[0117] Using a nonlinear optimizer, by adjusting θ i The optimal prediction model is obtained by minimizing the prediction error function of the total flow of the pumping station.
[0118] In practice, a pumping station has ten pumps operating in parallel. It is known that pumps numbered 1-3 are model A fixed-speed pumps, pumps numbered 4-6 are model B variable-frequency pumps, and pumps numbered 7-10 are model C fixed-speed pumps. Under normal circumstances, only one variable-speed pump is operated. The monitoring data includes the outlet pressure of each pump, the speed ratio of the variable-speed pump, the total pressure of the water leaving the pumping station, and the total flow rate.
[0119] In step A, the rated power, speed, head, and flow range of the three pump models were collected, and the factory-issued pump characteristic curve data were read. Relevant data were read from the SCADA system, and following step A4, the outlet pressure, speed ratio (for variable frequency pumps), factory-issued total head, and flow rate of each pump were compiled.
[0120] In step B1, characteristic curves of pumps based on the similarity principle are established. The coefficients of the characteristic curves for the three types of pumps are as follows, and the fitting results are shown in Table 1 below:
[0121] Table 1 Fitting Results
[0122] model Hx Sx A 61.7 -6.95e-07 B 65.4 -6.75e-07 C 54.2 -7.40e-06
[0123] In step B2, the error in the total factory flow rate mainly comes from the calculation error of the variable frequency pump flow rate. The further the operating frequency of the variable frequency pump is from the rated frequency, the greater the similarity theory error. Therefore, the key to correcting the characteristic curve lies in correcting the fitting formula of the variable frequency pump. Calculate the flow rate of each pump according to the first model in B1, and then compare it with the actual total factory flow rate data. Select the operating time of a single variable frequency pump (in this example, select pump #4), and use a Gaussian regression model to establish the relationship between the difference and the pump head H and speed ratio S. Table 2 below shows the data:
[0124] Table 2 Data Illustration Table
[0125]
[0126]
[0127] The parameters of the Gaussian regression model are shown in Table 3:
[0128] Table 3. Parameters of Gaussian Regression Model
[0129]
[0130] In step C, the flow rate is calculated based on the head and speed ratio of pump #4, and the relevant errors are effectively eliminated. For example... Figure 2 The figures show the actual flow rate of pump #4, the flow rate predicted by the curve fitting model (step B1 method), and the flow rate predicted by the final fusion model.
[0131] This invention overcomes the limitations of existing technologies in the application scenarios of fitting water pump characteristic curves and the instability and unreliability of calculation results; this invention can be carried out under normal water supply conditions without affecting the normal operation of the pumping station; when the water supply conditions change, this invention can still provide good calculation results.
[0132] Example 2
[0133] like Figure 3As shown, the difference between this embodiment and Embodiment 1 is that this embodiment provides a pump characteristic curve fitting system based on multi-source data fusion. This system uses a pump characteristic curve fitting method based on multi-source data fusion from Embodiment 1. This system corresponds one-to-one with the pump characteristic curve fitting method based on multi-source data fusion from Embodiment 1. The system includes:
[0134] The acquisition unit is used to acquire multi-source pumping station data and preprocess the multi-source pumping station data to obtain preprocessed multi-source pumping station data.
[0135] The model building unit includes a first model building subunit and a second model building subunit. The first model building subunit is a pump characteristic curve fitting model based on the similarity principle. It is constructed based on the factory test data of the pump station. The second model building subunit is constructed based on the pre-processed historical data of the pump station. The second model is the relationship between the actual flow difference and the head and speed ratio of the variable frequency pump.
[0136] The fusion model forming unit is used to fuse the first model and the second model to form a multi-source data fusion model; and the final prediction result is obtained based on the multi-source data fusion model.
[0137] The solver unit is used to optimize the parameters of the multi-source data fusion model. By conserving the total flow, the parameters of the second model are adjusted to obtain the optimal fitting result.
[0138] As a further implementation, the execution process of the first model construction sub-unit is as follows:
[0139] Based on the characteristic curve equation of a fixed-frequency pump, the expression of the characteristic curve equation for the pump's fixed-frequency operation is determined.
[0140] Based on the proportional law of similarity theory, the characteristic curve equation of the water pump under constant frequency operation is rewritten to obtain the characteristic curve expression of the water pump under variable frequency regulation.
[0141] Based on the experimental data before delivery, the characteristic curve expression of the water pump under variable frequency regulation is fitted to form the first model; based on the first model, the first predicted flow rate is calculated.
[0142] As a further implementation, the execution process of the second model construction sub-unit is as follows:
[0143] Based on the first model, calculate the first predicted flow rate of the variable frequency pump, and the flow rate difference between the actual flow rate of the variable frequency pump and the first predicted flow rate;
[0144] The relationship between the flow difference and the head and speed ratio of the variable frequency pump is established based on the Gaussian process model, forming a second model; the second predicted flow rate is calculated based on the second model.
[0145] As a further implementation, the execution process of fusing the model forming unit and the solution unit is as follows:
[0146] The calculation results of the first model and the second model are added together to form a multi-source data fusion model (i.e., a traffic prediction fusion model);
[0147] Based on actual historical operating data, the parameters of the second model are adjusted to minimize the total flow prediction error, and the optimal parameters are obtained.
[0148] The execution process of each unit can be carried out according to the process steps of the pump characteristic curve fitting method based on multi-source data fusion in Embodiment 1, and will not be described in detail in this embodiment.
[0149] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0150] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0151] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0152] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0153] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for fitting pump characteristic curves based on multi-source data fusion, characterized in that, The method includes: Acquire multi-source pump station data and preprocess the multi-source pump station data to obtain preprocessed multi-source pump station data; Based on the preprocessed multi-source pump station data, a first model and a second model are constructed. The first model is a pump characteristic curve fitting model based on the similarity principle, and the second model is the relationship between the flow difference and the head and speed ratio of the variable frequency pump. A multi-source data fusion model is formed by fusing the first and second models. The optimal parameters of the multi-source data fusion model are solved to obtain the optimal fitting result; The formula for calculating the first predicted flow rate is: ; in, The first predicted flow rate for the variable frequency pump; S and S represent the head and speed ratio of the variable frequency pump, respectively. and are the fitting coefficients of the variable frequency pump characteristic curve equation, and represent the maximum head of the device and the pipeline resistance coefficient, respectively; The formula for calculating the second predicted flow rate is: ; in, The second predicted flow rate for the variable frequency pump; GPR is a Gaussian prediction model; S and S represent the head and speed ratio of the variable frequency pump, respectively. These are the parameters for the Gaussian prediction model; The expression for the multi-source data fusion model is: ; in, This represents the final predicted flow rate of the variable frequency pump.
2. The method for fitting pump characteristic curves based on multi-source data fusion according to claim 1, characterized in that, The multi-source pump station data includes static pump data, pre-delivery test data, and historical operating data. The static data of the water pump includes the water pump model, water pump type, water pump rated power, and water pump rated speed; The pre-delivery test data refers to the pump station characteristic curve under standard operating conditions. The historical operating data includes the current status of the water pump and other key operating parameters. The current status of the water pump includes the pump's start-up and shutdown status and any modifications made. Other key operating parameters include the pump's outlet pressure, flow rate, and energy consumption.
3. The method for fitting pump characteristic curves based on multi-source data fusion according to claim 1, characterized in that, Based on the preprocessed multi-source pumping station data, a first model is constructed, including: Based on the characteristic curve equation of a fixed-frequency pump, the expression of the characteristic curve equation for the pump's fixed-frequency operation is determined. Based on the proportional law of similarity theory, the characteristic curve equation of the water pump under constant frequency operation is rewritten to obtain the characteristic curve expression of the water pump under variable frequency regulation. Based on the experimental data before delivery, the characteristic curve expression of the water pump under frequency conversion regulation is fitted to form a first model; based on the first model, the first predicted flow rate is calculated.
4. The method for fitting pump characteristic curves based on multi-source data fusion according to claim 1, characterized in that, Based on the preprocessed multi-source pumping station data, a second model is constructed, including: Based on the first model, calculate the first predicted flow rate of the variable frequency pump, and the flow rate difference between the actual flow rate of the variable frequency pump and the first predicted flow rate; The relationship between the flow difference and the head and speed ratio of the variable frequency pump is established based on the Gaussian process model to form a second model; the second predicted flow rate is calculated based on the second model.
5. The method for fitting pump characteristic curves based on multi-source data fusion according to claim 1, characterized in that, The optimal parameters of the multi-source data fusion model are solved by taking the final predicted flow rate of the variable frequency pump as a constraint and solving the optimal parameters of the objective function of the multi-source data fusion model. The objective function of the multi-source data fusion model is: ; The constraints are as follows: ; in, The total flow rate of the pumping station. To calculate the predicted flow rate of the i-th pump using the first and second models, Let be the parameters of the Gaussian process model for the i-th pump.
6. A pump characteristic curve fitting system based on multi-source data fusion, characterized in that, The system uses a pump characteristic curve fitting method based on multi-source data fusion as described in any one of claims 1 to 5; the system includes: The acquisition unit is used to acquire multi-source pumping station data and preprocess the multi-source pumping station data to obtain preprocessed multi-source pumping station data. The model building unit includes a first model building subunit and a second model building subunit. The first model building subunit is used to build a first model based on the preprocessed multi-source pump station data. The first model is a pump characteristic curve fitting model based on the similarity principle. The second model building subunit is used to build a second model based on the preprocessed multi-source pump station data. The second model is the relationship between the flow difference and the head and speed ratio of the variable frequency pump. A fusion model forming unit is used to fuse the first model and the second model to form a multi-source data fusion model. The solution unit is used to solve for the optimal parameters of the multi-source data fusion model and obtain the optimal fitting result.
7. A pump characteristic curve fitting system based on multi-source data fusion according to claim 6, characterized in that, The execution process of the first model construction subunit is as follows: Based on the characteristic curve equation of a fixed-frequency pump, the expression of the characteristic curve equation for the pump's fixed-frequency operation is determined. Based on the proportional law of similarity theory, the characteristic curve equation of the water pump under constant frequency operation is rewritten to obtain the characteristic curve expression of the water pump under variable frequency regulation. Based on the pre-delivery test data, the characteristic curve expression of the water pump under variable frequency regulation is fitted to form a first model; based on the first model, the first predicted flow rate is calculated. The execution process of the second model construction sub-unit is as follows: Based on the first model, calculate the first predicted flow rate of the variable frequency pump, and the flow rate difference between the actual flow rate of the variable frequency pump and the first predicted flow rate; The relationship between the flow difference and the head and speed ratio of the variable frequency pump is established based on the Gaussian process model to form a second model; the second predicted flow rate is calculated based on the second model. The execution process of the fusion model forming unit and the solution unit is as follows: The calculation results of the first model and the second model are added together to form a multi-source data fusion model; Based on actual historical operating data, the parameters of the second model are adjusted to minimize the total flow prediction error, and the optimal parameters are obtained.
Citation Information
Patent Citations
Method for calibrating performance curve of single water pump in urban water plant pump station
CN103062076A
Centrifugal pump external characteristic prediction method based on Gaussian process regression
CN112149913A
