Pump performance curve generation method, model building and training method and device

Abstract

The present invention discloses a method for generating a pump performance curve, a model building and training method and a device. The method includes: for each water pump to be tested in the pump room, obtaining the to-be-processed operation data corresponding to the current water pump to be tested and the to-be-processed liquid level data corresponding to the target water suction well associated with the current water pump to be tested; obtaining the preset water pump frequency and preset water pump flow range of the current water pump to be tested at the next moment of the current moment; processing the to-be-processed operation data, the to-be-processed liquid level data, the preset water pump frequency and the preset water pump flow range based on the pre-retrieved water pump pressure prediction sub-model to obtain the predicted water pump pressure range corresponding to the preset water pump flow range; based on the preset water pump flow range and the predicted water pump pressure range, determining the pump performance curve of the current water pump to be tested at the next moment. This technical solution increases the diversity and richness of the water pump performance analysis data, and thus improves the accuracy of the pump performance curve.

Description

Technical Field

[0001] The present invention relates to the technical field of water pump performance detection, and in particular to a pump performance curve generating method, model building and training method and device. Background Art

[0002] A water pump is a fluid machine that transports liquids or pressurizes liquids. It is widely used in petrochemical, nuclear power, irrigation, urban water supply and heating systems. The basic performance parameters of a water pump include flow, head, power, efficiency and speed. The curve that represents the relationship and change law between the main performance parameters of a water pump is usually called the performance curve of the water pump.

[0003] In related technologies, the pump holding method is usually used to generate a water pump performance curve. Specifically, the pump holding method is used to gradually close the valve and record the flow, head, power and other data of the water pump to draw the water pump performance curve. However, in daily operation, the pump holding method has limitations in terms of safety, economy and actual working conditions. Closing the valve of the water outlet pipe will cause the system pressure to increase, which may cause damage to the pipes and equipment, and increase energy consumption and maintenance costs. In addition, the pump holding method can only provide a performance curve in a closed state, and cannot accurately predict the performance of the water pump in actual working conditions. Summary of the invention

[0004] The present invention provides a pump performance curve generation method, model construction and training method and device, so as to achieve the effect of simulating and generating pump room operation data at a predicted time based on the historical operation data of the water pump, thereby increasing the diversity and richness of the water pump performance analysis data, and further improving the accuracy of the pump performance curve.

[0005] According to a first aspect of the present invention, a method for generating a pump performance curve is provided, the method comprising:

[0006] For each water pump to be tested in the pump room, obtain the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested; wherein the pending operation data include water pump frequency, water pump pressure and water pump flow; the pending operation data include current operation data at the current moment and historical operation data within a preset time period before the current moment; the pending liquid level data include current liquid level data at the current moment and historical liquid level data within the preset time period before the current moment;

[0007] Obtaining a preset water pump frequency and a preset water pump flow range of the current water pump to be tested at a moment next to the current moment;

[0008] Based on the pre-retrieved water pump pressure prediction sub-model, the to-be-processed operating data, the to-be-processed liquid level data, the preset water pump frequency and the preset water pump flow interval are processed to obtain a predicted water pump pressure interval corresponding to the preset water pump flow interval; wherein the water pump pressure prediction sub-model is a sub-model in the environmental model, the environmental model is constructed by at least one sub-model, and the at least one sub-model is spliced ​​according to a preset splicing order;

[0009] Based on the preset water pump flow range and the predicted water pump pressure range, a pump performance curve of the current water pump to be tested at the next moment is determined.

[0010] According to a second aspect of the present invention, there is provided a model building method, the method comprising:

[0011] Get multiple sub-models to be spliced;

[0012] The sub-models to be spliced ​​are spliced ​​together according to a preset splicing order to obtain a model to be trained.

[0013] According to a third aspect of the present invention, a model training method is provided, the method comprising:

[0014] Acquire multiple training samples; wherein the training samples include sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump flow and actual water pump pressure corresponding to the prediction time, sample pipeline flow, sample pipeline pressure, actual pipeline flow and actual pipeline pressure corresponding to the prediction time, and sample water absorption well liquid level;

[0015] For each training sample, the current training sample is input into the model to be trained to obtain the output results corresponding to each sub-model to be spliced;

[0016] Based on the output result, the model parameters in the corresponding sub-models to be spliced ​​are corrected, and the convergence of the loss functions in each of the sub-models to be spliced ​​is used as a training target to obtain an environmental model, so as to predict the water pump pressure of the water pump to be tested in the pump room under different preset water pump flow rates at the next moment based on the water pump pressure prediction sub-model in the environmental model to obtain the predicted water pump pressure.

[0017] According to a fourth aspect of the present invention, there is provided a pump performance curve generating device, the device comprising:

[0018] A data acquisition module is used to obtain, for each water pump to be tested in the pump room, the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested; wherein the pending operation data include water pump frequency, water pump pressure and water pump flow; the pending operation data include current operation data at the current moment and historical operation data within a preset time period before the current moment; the pending liquid level data include current liquid level data at the current moment and historical liquid level data within the preset time period before the current moment;

[0019] A preset data acquisition module, used to obtain a preset water pump frequency and a preset water pump flow range of the current water pump to be tested at a moment next to the current moment;

[0020] A data processing module, used for processing the to-be-processed operating data, the to-be-processed liquid level data, the preset water pump frequency and the preset water pump flow interval based on a pre-retrieved water pump pressure prediction sub-model, to obtain a predicted water pump pressure interval corresponding to the preset water pump flow interval; wherein the water pump pressure prediction sub-model is a sub-model in the environmental model, the environmental model is constructed by at least one sub-model, and the at least one sub-model is spliced ​​according to a preset splicing order;

[0021] The performance curve determination module is used to determine the pump performance curve of the current water pump to be tested at the next moment based on the preset water pump flow interval and the predicted water pump pressure interval.

[0022] According to a fifth aspect of the present invention, there is provided a model building device, the device comprising:

[0023] A sub-model acquisition module is used to acquire multiple sub-models to be spliced;

[0024] The sub-model splicing module is used to splice the sub-models to be spliced ​​together according to a preset splicing order to obtain a model to be trained.

[0025] According to a sixth aspect of the present invention, there is provided a model training device, the device comprising:

[0026] A training sample acquisition module, used to acquire multiple training samples; wherein the training samples include sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump flow and actual water pump pressure corresponding to the prediction time, sample pipeline flow, sample pipeline pressure, actual pipeline flow and actual pipeline pressure corresponding to the prediction time, and sample water absorption well liquid level;

[0027] The output result determination module is used to input the current training sample into the model to be trained for each training sample, and obtain the output result corresponding to each sub-model to be spliced;

[0028] A model parameter correction module is used to correct the model parameters in the corresponding sub-models to be spliced ​​based on the output results, and take the convergence of the loss functions in each of the sub-models to be spliced ​​as a training target to obtain an environmental model, so as to predict the water pump pressure of the water pump to be tested in the pump room at different preset water pump flow rates at the next moment based on the water pump pressure prediction sub-model in the environmental model to obtain a predicted water pump pressure.

[0029] According to a seventh aspect of the present invention, there is provided an electronic device, the electronic device comprising:

[0030] at least one processor; and

[0031] a memory communicatively connected to the at least one processor; wherein,

[0032] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the pump performance curve generating method, model building and training method described in any embodiment of the present invention.

[0033] According to an eighth aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the pump performance curve generation method, model construction and training method described in any embodiment of the present invention when executed.

[0034] The technical solution of the embodiment of the present invention obtains the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested for each water pump to be tested in the pump room; then, obtains the preset water pump frequency and preset water pump flow range of the current water pump to be tested at the next moment of the current moment; further, based on the pre-retrieved water pump pressure prediction sub-model, the pending operation data, the pending liquid level data, the preset water pump frequency and the preset water pump flow range are processed to obtain the predicted water pump pressure range corresponding to the preset water pump flow range; finally, based on the preset water pump flow range and the predicted water pump pressure range, the pump performance curve of the current water pump to be tested at the next moment is determined, which solves the problems of increasing energy consumption and maintenance costs when generating pump performance curves in related technologies, failing to accurately predict the performance of the water pump in actual working conditions, and failing to cover all possible working conditions, and realizes the effect of simulating and generating pump room operation data at the predicted moment based on the historical operation data of the water pump, increases the diversity and richness of the water pump performance analysis data, thereby improving the accuracy of the pump performance curve, and realizing the effect of real-time reflection of the dynamic changes of the pump room system.

[0035] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 is a flow chart of a method for generating a pump performance curve according to Embodiment 1 of the present invention;

[0038] Figure 2 is a flow chart of a model building method provided according to Embodiment 2 of the present invention;

[0039] Figure 3 is a flowchart of a model training method provided according to Embodiment 3 of the present invention;

[0040] Figure 4 is a schematic diagram of a model training process provided according to Embodiment 3 of the present invention;

[0041] Figure 5 is a schematic structural diagram of a pump performance curve generating device provided according to a fourth embodiment of the present invention;

[0042] Figure 6 is a structural schematic diagram of a model building device provided according to Embodiment 5 of the present invention;

[0043] Figure 7 is a structural schematic diagram of a model training device provided according to Embodiment 6 of the present invention;

[0044] Figure 8 It is a structural schematic diagram of an electronic device for implementing the pump performance curve generating method, model building and training method of an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0046] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0047] Embodiment 1

[0048] Figure 1 This is a flow chart of a method for generating a pump performance curve provided by the first embodiment of the present invention. This embodiment is applicable to the case of detecting the operating performance of any water pump in a pump room. The method can be executed by a pump performance curve generating device. The pump performance curve generating device can be implemented in the form of hardware and / or software. The pump performance curve generating device can be configured in a terminal and / or a server. Figure 1 As shown, the method includes:

[0049] S110 . For each water pump to be tested in the pump room, obtain the to-be-processed operation data corresponding to the current water pump to be tested and the to-be-processed liquid level data corresponding to the target water suction well.

[0050] Among them, the pump room is a plant for installing water pumps, power machines and their auxiliary equipment. In this embodiment, the pump room may include multiple water pumps, multiple water suction wells and multiple sets of liquid transmission pipelines and other equipment. The water pump to be tested can be understood as the water pump that needs to be tested in the pump room. The water pump to be tested can be all the water pumps in the pump room, or it can be at least one water pump in the pump room. Optionally, the water pump to be tested is a centrifugal pump. The operation data to be processed can be understood as the operation data associated with various water pump performance parameters generated during the operation of the water pump to be tested. The operation data to be processed can include water pump frequency, water pump pressure and water pump flow. It can be understood by those skilled in the art that the water pump frequency refers to the frequency of the water pump speed, which is usually expressed in Hertz (Hz). The size of the water pump frequency can affect the working effect, life and power consumption of the water pump. The water pump pressure refers to the force that the water pump needs to apply to the water flow when working. The water pump pressure can also be understood as the water pump outlet pressure, that is, the water pressure at the water pump outlet. The size of the water pump pressure can affect the working efficiency of the water pump and the flow rate of the water flow. There is a positive correlation between water pump pressure and water pump head. The water pump head can be determined by the water pump pressure, and correspondingly, the water pump pressure can also be determined by the water pump head. The water pump flow rate refers to the amount of liquid (liquid volume or liquid weight) delivered by the pump per unit time. It should be noted that the operating data to be processed may also include other operating data generated during the operation of the pump room, which is not specifically limited in this embodiment.

[0051] In this embodiment, the operation data to be processed includes the current operation data at the current moment and the historical operation data within a preset time before the current moment. The preset time can be a time interval of any length. Optionally, the preset time can be 10 minutes, 20 minutes, or 30 minutes, etc.

[0052] Among them, the target suction well is the suction well associated with the current water pump to be tested, that is, the suction well associated with the current water pump to be tested among all the suction wells included in the pump room. Generally, a pump room can include a preset number of suction wells. The preset number can be any value, optionally, it can be 2 or 3. Those skilled in the art will understand that the suction well refers to a structure specially set up for the water pump suction pipe. The suction well liquid level data is the data that characterizes the height of the liquid in the suction well. In this embodiment, the liquid level data to be processed includes the current liquid level data at the current moment and the historical liquid level data within a preset time period before the current moment period.

[0053] In practical applications, during the operation of each water pump in the pump room, for each water pump, the water pump frequency generated by the current water pump during operation can be detected based on the frequency detection device, and the detected horizontal frequency can be sampled according to the preset sampling period. Then, the water pump frequency corresponding to each sampling point during the operation of the current water pump can be obtained and stored in the database of the corresponding terminal device. At the same time, the water pump pressure generated by the current water pump during operation can also be detected based on the pressure detection device, and the detected water pump pressure can be sampled according to the preset sampling period. Then, the water pump pressure corresponding to each sampling point during the operation of the current water pump can be obtained and stored in the database of the corresponding terminal device. At the same time, the water pump flow generated by the current water pump during operation can also be detected based on the flow detection device, and the detected water pump flow can be sampled according to the preset sampling period. Then, the water pump flow corresponding to each sampling point during the operation of the current water pump can be obtained and stored in the database of the corresponding terminal device. Thus, the operation data corresponding to each water pump in the pump room can be obtained. At the same time, during the operation of each water pump in the pump room, for each water suction well associated with each water pump, the water suction well liquid level generated during the operation of the current water suction well can be detected based on the liquid level detection equipment, and the detected water suction well liquid level can be sampled according to the preset sampling period. Then, the water suction well liquid level data corresponding to each sampling point during the operation of the current water suction well can be obtained and stored in the database of the corresponding terminal device. Among them, the preset sampling period can be any value, and can be 1 second or 2 seconds, etc.

[0054] Further, in order to detect the performance of each water pump to be tested in the pump room, for each water pump to be tested in the pump room, the historical operation data of the current water pump to be tested within the preset time before the current moment can be retrieved from the database, and the operation data can include at least one of the water pump frequency, water pump pressure and water pump flow. In addition, the current operation data of the current water pump to be tested at the current moment can also be obtained, and the current operation data can also include at least one of the water pump frequency, water pump pressure and water pump flow. Then, the current operation data and the historical operation data can be used as the operation data to be processed to obtain the operation data to be processed corresponding to the current water pump to be tested. Further, in the case of determining the current water pump to be tested, the target water suction well associated with the current water pump to be tested can be determined. Then, the historical liquid level data of the target water suction well within the preset time before the current moment can be retrieved from the database, and the current liquid level data of the target water suction well at the current moment can be obtained. Then, the historical liquid level data and the current liquid level data can be used as the liquid level data to be processed to obtain the liquid level data to be processed corresponding to the target water suction well.

[0055] S120, obtaining a preset water pump frequency and a preset water pump flow range of the water pump to be tested at the next moment after the current moment.

[0056] The preset water pump frequency can be understood as a preset water pump frequency. The preset water pump frequency can be a water pump frequency that meets the water pump performance detection requirements. The preset water pump flow interval includes multiple preset water pump flow rates, and the multiple preset water pump flow rates are obtained based on the preset step size. The preset water pump flow interval can be understood as a preset water pump flow range, and the flow range can be a range that meets the water pump performance detection requirements.

[0057] Generally speaking, the water pump performance detection of a centrifugal pump can be achieved by determining the relationship between the water pump flow rate and the water pump head of the water pump. At the same time, the water pump head can be determined based on the water pump pressure. Therefore, the water pump pressure corresponding to different water pump flow rates of the water pump at any time can be determined, and the water pump head can be determined based on the water pump pressure. Furthermore, a pump performance curve can be generated based on multiple water pump flow rates and their corresponding water pump heads to analyze the water pump performance of the water pump based on the pump performance curve.

[0058] For the technical solution provided by this implementation, in the actual application process, the pump frequency, pump pressure, pump flow rate and water suction well liquid level data of the water pump to be tested at the current moment and within the preset time length before the current moment can be obtained. Further, in order to obtain the water pump pressure corresponding to different water pump flow rates of the water pump to be tested at the next moment, the water pump frequency at the next moment and the water pump flow range including different water pump flow rates can be set to obtain the preset water pump frequency and preset water pump flow range of the water pump to be tested at the next moment. Furthermore, the water pump frequency, pump pressure, pump flow rate and water suction well liquid level data of the water pump to be tested at the current moment and within the preset time length before the current moment can be processed to determine the mapping relationship between the water pump frequency, water pump flow rate and water pump pressure of the water pump to be tested. Afterwards, the preset water pump frequency and preset water pump flow range of the next moment can be processed according to the mapping relationship, so that the water pump pressure corresponding to different preset water pump flow rates of the water pump to be tested at the next moment can be obtained.

[0059] S130, based on the pre-retrieved water pump pressure prediction sub-model, the to-be-processed operating data, the to-be-processed liquid level data, the preset water pump frequency and the preset water pump flow range are processed to obtain a predicted water pump pressure range corresponding to the preset water pump flow range.

[0060] The water pump pressure prediction sub-model may be a pre-trained neural network model that takes the water pump operation data and the liquid level data of the water suction well as input objects to predict the water pump pressure after the current moment based on the input objects. In this embodiment, the water pump pressure prediction sub-model is a sub-model in the environmental model, and the environmental model is constructed by at least one sub-model, and the at least one sub-model is spliced ​​according to a preset splicing order.

[0061] Among them, the environmental model can be understood as a model that characterizes the operation of the pump room, or it can be understood as a model obtained after modeling the real operating environment of the pump room. The environmental model can be a modeling method in the field of reinforcement learning. Modeling the environment is for planning, that is, to obtain a better strategy. The so-called planning is to take the model as input, and then, by interacting with the environmental model, generate or improve the calculation process of a strategy. The environmental model may include a state transfer function (state transfer model) and a reward function. For this embodiment, the environmental model may be a model established for better planning of the operation of the pump room. Generally, the pump room may include a water pump, a pipeline and a water suction well. Among them, the water pump and the pipeline may affect the operation of the pump room due to changes in their performance, while the water suction well does not have the problem of performance changes. Therefore, the environmental model corresponding to the pump room may include at least one sub-model, and at least one sub-model may be spliced ​​in a preset order. Optionally, at least one sub-model may include a water pump pressure prediction sub-model, a water pump flow prediction sub-model, a pipeline flow prediction sub-model, and a pipeline pressure prediction sub-model. The preset splicing order can be explained by the model input and output between each sub-model. Exemplarily, the model output of the water pump flow prediction submodel can be the model input of the water pump pressure prediction submodel and the pipeline flow prediction submodel, and the model output of the water pump pressure prediction submodel and the pipeline flow prediction submodel can be the model input of the pipeline pressure prediction submodel.

[0062] Among them, each sub-model can be a state transition model, that is, the water pump pressure prediction model is a state transition model. Those skilled in the art can understand that the state transition model can be understood as an N-order Markov model, and the N-order Markov model can be obtained after modeling the Markov process. The Markov process refers to the transfer of each state in the process only depends on the previous N states. For the water pump pressure prediction model provided in this embodiment, the data set consisting of the operation data to be processed, the liquid level data to be processed, the preset water pump frequency and the preset water pump flow interval can be used as the state space, and the corresponding relationship between the water pump frequency, the water pump flow and the water pump pressure can be used as the state transition relationship. In the actual application process, after the state space is input into the water pump pressure prediction model, the operation data to be processed and the liquid level data to be processed can be processed based on the water pump pressure prediction model, the corresponding relationship between the water pump frequency, the water pump flow and the water pump pressure of the current water pump to be tested is determined, and the state transition relationship is constructed, and then, the preset water pump frequency and multiple preset water pump flows can be processed according to the state transition relationship, so that the predicted water pump pressure corresponding to different preset water pump flows at the preset water pump frequency can be obtained.

[0063] It should be noted that each sub-model in the environmental model provided in this embodiment is a neural network model composed of a generator and a discriminator. When training the model, the goal of the discriminator is to identify the state at the current moment as true and the generated state at the next moment as false; the goal of the generator is to make the generated state at the next moment gradually close to the real state at the next moment, so that the discriminator can identify the generated state as true. In the continuous training and optimization of the discriminator and the generator, the data generated by the generator can be closer and closer to the real data distribution, until the discriminator can no longer identify whether the state data is real historical data or predicted data, thereby ensuring the accuracy of the environmental model.

[0064] In practical applications, multiple training samples can be obtained. Furthermore, for each training sample, the current training sample can be input into the environmental model to be trained, and the output results corresponding to each sub-model in the environmental model can be obtained. Based on the output results, the model parameters in the corresponding sub-model are corrected, and the convergence of the loss function in each sub-model is used as the training goal to obtain the trained environmental model, that is, each trained sub-model can be obtained. Furthermore, in order to predict the water pump pressure of the water pump to be tested in the pump room at the next moment, the water pump pressure prediction sub-model can be retrieved from the trained environmental model.

[0065] Furthermore, the acquired processing operation data corresponding to the current water pump to be tested, the processing liquid level data corresponding to the target water suction well, the preset water pump frequency at the next moment and the preset water pump flow range can be input into the water pump pressure prediction sub-model, so as to process the input data based on the water pump pressure prediction sub-model to obtain the predicted water pump pressure corresponding to each preset water pump flow.

[0066] Optionally, the operating data to be processed, the liquid level data to be processed, the preset water pump frequency and the preset water pump flow interval are processed based on a pre-retrieved water pump pressure prediction sub-model to obtain a predicted water pump pressure range corresponding to the preset water pump flow interval, including: dividing the preset water pump flow interval based on a pre-division step size to obtain multiple preset water pump flows, and constructing a group of data to be processed based on the operating data to be processed, the liquid level data to be processed, the preset water pump frequency and the preset water pump flow to obtain multiple groups of data to be processed; processing the multiple groups of data to be processed based on the water pump pressure prediction sub-model to obtain the predicted water pump pressure corresponding to each group of data to be processed; and determining the predicted water pump pressure range based on the predicted water pump pressures corresponding to the multiple groups of data to be processed.

[0067] The preset step size can be understood as a preset water pump flow division step size. The preset step size can be any value, and can be 10 liters / second, 50 liters / second, or 100 liters / second. For example, assuming that the preset water pump flow interval is 4000 liters / second-8000 liters / second, and the preset step size is 10 liters / second, the preset water pump flow interval is divided according to the preset step size, and 400 preset water pump flows can be obtained.

[0068] In practical applications, the preset water pump flow interval can be divided according to the preset division step size to obtain multiple preset water pump flows. Furthermore, the data composed of the operation data to be processed, the liquid level data to be processed, the preset water pump frequency and the single preset water pump flow can be used as a group of data to be processed, so as to obtain multiple groups of data to be processed. Further, the multiple groups of data to be processed can be processed based on the water pump pressure prediction submodel to obtain the predicted water pump pressure corresponding to each group of data to be processed. Specifically, after the multiple groups of data to be processed are input into the water pump pressure prediction submodel, for each group of data, the operation data to be processed and the liquid level data to be processed can be processed based on the water pump pressure prediction submodel to determine the corresponding relationship between the water pump frequency, the water pump pressure and the water pump flow of the current water pump to be tested at the current moment and within the preset time before the current moment. Furthermore, the preset water pump frequency and the preset water pump flow in the current group of data to be processed can be analyzed according to the corresponding relationship to obtain the predicted water pump pressure corresponding to the preset water pump flow at the preset water pump frequency. Furthermore, after obtaining the predicted water pump pressures corresponding to multiple groups of data to be processed, each predicted water pump pressure can be sorted according to the arrangement order of the corresponding preset water pump flow in the preset water pump flow range to obtain the predicted water pump pressure range corresponding to the preset water pump flow range.

[0069] S140: Determine a pump performance curve of the current water pump to be tested at the next moment based on a preset water pump flow range and a predicted water pump pressure range.

[0070] Among them, the pump performance curve can be a curve reflecting the relationship between various performance parameters of the water pump. In general, it is extremely important to fully understand the performance of the water pump, be familiar with the characteristics of the performance curve, and master its changing rules for reasonable selection and matching, correctly determine the installation height of the water pump, adjust the operating conditions of the water pump, and strengthen the scientific management of the pump station. The pump performance curve refers to the relationship between the pump performance parameters in the form of a curve. Usually, the horizontal axis represents the water pump flow, and the vertical axis can represent the water pump head, water pump efficiency or water pump shaft power, etc., which are respectively called flow-head curve, flow-efficiency curve or flow-shaft power curve. In this embodiment, the pump performance curve can be a flow-head curve, that is, a curve reflecting the corresponding relationship between the water pump flow and the water pump head. It should be noted that in the field of water pump technology, there is a certain conversion relationship between the water pump pressure and the water pump head. When the water pump pressure is obtained, the corresponding water pump head can be determined according to the conversion relationship. Therefore, after obtaining the predicted water pump pressure range, the predicted water pump head range corresponding to the predicted water pump pressure range can be determined according to the conversion relationship between the water pump pressure and the water pump head. Furthermore, the pump performance curve of the current water pump to be tested at the next moment can be determined based on the preset water pump flow range and the predicted water pump head range.

[0071] Optionally, based on a preset water pump flow interval and a predicted water pump pressure range, a pump performance curve of the current water pump to be tested at the next moment is determined, including: determining a predicted water pump head range corresponding to the predicted water pump pressure range; based on the preset water pump flow interval and the predicted water pump head range, generating a pump performance curve of the current water pump to be tested at the next moment.

[0072] The predicted pump head range can be understood as a data set consisting of multiple continuous predicted pump heads. Those skilled in the art can understand that the pump head refers to the height to which the pump can lift water, that is, the highest point to which the pump can deliver water from the water inlet. There is a positive correlation between the pump head and the pump pressure. For example, a pump pressure of 1 MPa can be approximately equal to a pump head of 100 meters.

[0073] In practical applications, after obtaining the predicted water pump pressure range, for each predicted water pump pressure in the predicted water pump pressure range, the current predicted water pump pressure can be processed according to a predetermined conversion formula, and then the predicted water pump head corresponding to the current predicted water pump pressure can be obtained.

[0074] Exemplarily, the predicted water pump head corresponding to the current predicted water pump pressure may be determined based on the following formula:

[0075]

[0076] Where H can represent the predicted pump head corresponding to the current predicted pump pressure; P 出It can represent the current predicted water pump pressure; P 入 It can represent the inlet pressure of the current water pump to be tested, which can be determined based on the liquid level data to be processed; ρ can represent the liquid density; g represents the gravity coefficient; h can represent the distance between the liquid level of the target water suction well and the suction port of the current water pump to be tested.

[0077] Furthermore, after obtaining the predicted water pump head range corresponding to the predicted water pump pressure range, the corresponding relationship between the preset water pump flow and the predicted water pump head can be determined based on the preset water pump flow interval and the predicted water pump head range. Furthermore, a curve reflecting the corresponding relationship can be generated with the preset water pump flow as the horizontal coordinate and the predicted water pump head as the vertical coordinate, so as to obtain the pump performance curve of the current water pump to be tested at the next moment. Therefore, the water pump performance of the current water pump to be tested can be analyzed according to the pump performance curve, and the operating parameters of the current water pump to be tested can be adjusted based on the analysis results, so that the pump room always maintains a high operating efficiency during operation.

[0078] The technical solution of the embodiment of the present invention obtains the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested for each water pump to be tested in the pump room; then, obtains the preset water pump frequency and preset water pump flow range of the current water pump to be tested at the next moment of the current moment; further, based on the pre-retrieved water pump pressure prediction sub-model, the pending operation data, the pending liquid level data, the preset water pump frequency and the preset water pump flow range are processed to obtain the predicted water pump pressure range corresponding to the preset water pump flow range; finally, based on the preset water pump flow range and the predicted water pump pressure range, the pump performance curve of the current water pump to be tested at the next moment is determined, which solves the problems of increasing energy consumption and maintenance costs when generating pump performance curves in related technologies, failing to accurately predict the performance of the water pump in actual working conditions, and failing to cover all possible working conditions, and realizes the effect of simulating and generating pump room operation data at the predicted moment based on the historical operation data of the water pump, increases the diversity and richness of the water pump performance analysis data, thereby improving the accuracy of the pump performance curve, and realizing the effect of real-time reflection of the dynamic changes of the pump room system.

[0079] Embodiment 2

[0080] Figure 2 This is a flow chart of a model building method provided by the second embodiment of the present invention. This embodiment can be applied to modeling any environment to build a model that reflects the corresponding environmental changes. Optionally, an environmental model corresponding to the pump room environment is built. The method can be executed by a model building device, which can be implemented in the form of hardware and / or software, and can be configured in a terminal and / or server. Figure 2 As shown, the method includes:

[0081] S210: Acquire multiple sub-models to be spliced.

[0082] Among them, the sub-model to be spliced ​​can be understood as a pre-built model framework to be trained. The sub-model to be spliced ​​can be a neural network model of any model structure. Exemplarily, the sub-model to be spliced ​​can be a generative adversarial state transfer model, that is, a state transfer sub-model including a generator and a discriminator. In this embodiment, the multiple sub-models to be spliced ​​can include at least four sub-models to be spliced, and the at least four sub-models to be spliced ​​can include a water pump flow prediction sub-model to be spliced, a water pump pressure prediction sub-model to be spliced, a pipeline flow prediction sub-model to be spliced, and a pipeline pressure prediction sub-model to be spliced.

[0083] In practical applications, multiple model frameworks can be constructed according to a predetermined model structure, and then multiple sub-models to be spliced ​​can be obtained. Therefore, the sub-models to be spliced ​​can be spliced ​​together to obtain a model including multiple sub-models to be spliced.

[0084] S220, splicing the sub-models to be spliced ​​together according to a preset splicing order to obtain a model to be trained.

[0085] The preset splicing order may be a predetermined order for splicing the sub-models. The model to be trained may be a model including a plurality of sub-models to be spliced, and the model parameters are default values. In this embodiment, the model to be trained may be an environmental model to be trained. The model to be trained is trained with a plurality of training samples representing the operation of the pump room to obtain an environmental model, which may be used as a model corresponding to the operating environment of the pump room.

[0086] In practical applications, after obtaining multiple sub-models to be spliced, the sub-models to be spliced ​​can be spliced ​​together according to a preset splicing order, and the model obtained after splicing can be used as a model to be trained. It should be noted that the preset splicing order corresponding to each sub-model to be spliced ​​can be explained by the relationship between the model output and the model input of each sub-model to be spliced.

[0087] Optionally, the sub-models to be spliced ​​together in a preset splicing order to obtain the model to be trained, including: using the model input of the model to be trained as the model input of the water pump flow prediction sub-model to be spliced, and using the model output of the water pump flow prediction sub-model to be spliced ​​as at least part of the model input of the water pump pressure prediction sub-model to be spliced ​​and the pipeline flow prediction sub-model to be spliced; using the model output of the water pump pressure prediction sub-model to be spliced ​​and the model output of the pipeline flow prediction sub-model to be spliced ​​as at least part of the model input of the pipeline pressure prediction sub-model to be spliced, and using the model output of each sub-model to be spliced ​​as the model output of the model to be trained, so as to obtain the model to be trained.

[0088] The technical solution of the embodiment of the present invention obtains multiple sub-models to be spliced, and further splices the sub-models to be spliced ​​together according to a preset splicing order to obtain a model to be trained, thereby achieving the effect of constructing an environmental model to be trained corresponding to the operating conditions of the pump room. Furthermore, it can achieve the effect of real-time tracking of the pump room operating data and real-time export of the performance curve of the water pump based on the trained environmental model.

[0089] Embodiment 3

[0090] Figure 3 This is a schematic diagram of a model training method provided by the third embodiment of the present invention and applied to the model to be trained described in the above embodiment. After constructing a model to be trained including multiple sub-models to be spliced, the model to be trained can be trained based on training samples to obtain an environmental model. Thus, the operation of the pump room can be analyzed based on the environmental model. The specific implementation method can refer to the technical solution of this embodiment. Among them, the technical terms that are the same or similar to those in the above embodiment will not be described in detail here.

[0091] like Figure 3 As shown, the method includes:

[0092] S310: Obtain multiple training samples.

[0093] Among them, the training samples include sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump flow and actual water pump pressure corresponding to the prediction time, sample pipeline flow, sample pipeline pressure, actual pipeline flow and actual pipeline pressure corresponding to the prediction time and sample water absorption well liquid level.

[0094] In this embodiment, the sample water pump flow rate can be the water pump flow rate collection data of all water pumps in the pump room within a preset historical time length. The sample water pump pressure can be the water pump pressure collection data of all water pumps in the pump room within a preset historical time length. The sample water pump frequency can be the water pump frequency collection data of all water pumps in the pump room within a preset historical time length. It should be noted that the sample water pump flow rate, the sample water pump pressure and the sample water pump frequency can all be data pre-stored in the database. The prediction time can be any time within the historical operation time period, and the time is the time after the preset historical time length corresponding to the sample data. Exemplarily, assuming that the preset historical time length corresponding to the sample data can be (tn, t], the prediction time can be t+1. Moreover, (tn, t] and t+1 are both historical time intervals. The actual water pump flow rate can be the water pump flow rate collection data of all water pumps in the pump room at the predicted time. The actual water pump pressure can be the water pump pressure collection data of all water pumps in the pump room at the predicted time.

[0095] In this embodiment, the sample pipeline flow rate may be the pipeline flow rate data collected for all pipelines in the pump room within a preset historical time period. The sample pipeline pressure may be the pipeline pressure data collected for all pipelines in the pump room within a preset historical time period. The actual pipeline flow rate may be the pipeline flow rate data collected for all pipelines in the pump room at a predicted time. The actual pipeline pressure may be the pipeline pressure data collected for all pipelines in the pump room at a predicted time. The sample water absorption well liquid level may be the liquid level data collected for all water absorption wells in the pump room within a preset historical time period.

[0096] In practical applications, sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump flow and actual water pump pressure corresponding to the prediction time, sample pipeline flow, sample pipeline pressure, actual pipeline flow and actual pipeline pressure corresponding to the prediction time, and sample water well liquid level can be obtained from the database. Furthermore, training samples can be constructed based on the acquired pump room operation data to obtain multiple training samples.

[0097] S320: For each training sample, input the current training sample into the model to be trained to obtain output results corresponding to each sub-model to be spliced.

[0098] Among them, when the multiple sub-models to be spliced ​​in the model to be trained are the water pump flow prediction sub-model to be spliced, the water pump pressure prediction sub-model to be spliced, the pipeline flow prediction sub-model to be spliced ​​and the pipeline pressure prediction sub-model to be spliced, the output results may include the predicted water pump flow, the predicted water pump pressure, the predicted pipeline flow and the predicted pipeline pressure.

[0099] In practical applications, for each training sample, the current training sample can be input into the model to be trained, so as to process the current training sample based on each sub-model to be spliced ​​in the model to be trained, and then the output results corresponding to each sub-model to be spliced ​​can be obtained.

[0100] Optionally, the current training sample is input into the model to be trained to obtain the actual output results corresponding to each sub-model to be spliced, including: inputting the sample water pump flow, sample water pump frequency, sample water suction well liquid level and sample pipeline pressure in the current training sample into the water pump flow prediction sub-model to be spliced ​​in the model to be trained to obtain the predicted water pump flow corresponding to the current training sample; inputting the sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump frequency and predicted water pump flow into the water pump pressure prediction sub-model to be spliced ​​in the model to be trained to obtain the predicted water pump pressure corresponding to the current training sample; inputting the sample pipeline flow, sample water pump pressure and predicted water pump flow into the pipeline flow prediction sub-model to be spliced ​​in the model to be trained to obtain the predicted pipeline flow corresponding to the current training sample; inputting the sample pipeline flow, sample pipeline pressure, sample water pump pressure, predicted water pump pressure and predicted pipeline flow into the pipeline pressure prediction sub-model to be spliced ​​in the model to be trained to obtain the pipeline predicted pressure corresponding to the current training sample.

[0101] In practical applications, after the current training sample is input into the model to be trained, the sample water pump flow, sample water pump frequency, sample water well liquid level and sample pipeline pressure in the current training sample can be input into the water pump flow prediction sub-model to be spliced, so as to process the input data based on the water pump flow prediction sub-model to be spliced, and obtain the predicted water pump flow corresponding to the prediction time. Afterwards, the sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump frequency and predicted water pump flow are input into the water pump pressure prediction sub-model to be spliced, so as to process the input data based on the water pump pressure prediction sub-model to be spliced, and obtain the predicted water pump pressure corresponding to the prediction time. Afterwards, the sample pipeline flow, sample water pump pressure and predicted water pump flow can be input into the pipeline flow prediction sub-model to be spliced, so as to process the input data based on the pipeline flow prediction sub-model to be spliced, and obtain the predicted pipeline flow corresponding to the prediction time. Finally, the sample pipeline flow, sample pipeline pressure, sample water pump pressure, predicted water pump pressure and predicted pipeline flow are input into the pipeline pressure prediction sub-model to be spliced, so as to process the input data based on the pipeline pressure prediction sub-model to obtain the predicted pipeline pressure corresponding to the prediction time.

[0102] For example, Figure 4 As shown in the figure, it is a schematic diagram of the training process of the model to be trained. Figure 4It can be seen that the sample water pump flow, sample water pump frequency, sample water suction well liquid level and sample pipeline pressure in the training sample are input into the water pump flow prediction sub-model M110 to obtain the predicted water pump flow; then, the sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump frequency and predicted water pump flow are input into the water pump pressure prediction sub-model M120 to obtain the predicted water pump pressure; then, the sample pipeline flow, sample water pump pressure and predicted water pump flow are input into the pipeline flow prediction sub-model M130 to obtain the predicted pipeline flow; finally, the sample pipeline flow, sample pipeline pressure, sample water pump pressure, predicted water pump pressure and predicted pipeline flow are input into the pipeline pressure prediction sub-model M140 to obtain the predicted pipeline pressure.

[0103] S330. Based on the output results, the model parameters in the corresponding sub-models to be spliced ​​are corrected, and the convergence of the loss functions in each sub-model to be spliced ​​is used as the training target to obtain the environmental model, and the water pump pressure prediction sub-model in the environmental model is used to predict the water pump pressure of the water pump to be tested in the pump room under different preset water pump flow rates at the next moment to obtain the predicted water pump pressure.

[0104] The loss function may be a function determined based on the loss value and used to characterize the degree of difference between the actual output of the model and the theoretical output.

[0105] In practical applications, after obtaining the output results corresponding to each sub-model to be spliced, the model parameters in the corresponding sub-model to be spliced ​​can be corrected based on each output result. Then, the convergence of the loss function in each sub-model to be spliced ​​can be used as a training goal to obtain the environment model.

[0106] Optionally, based on the output result, the model parameters in the corresponding sub-models to be spliced ​​are corrected, and the convergence of the loss function in each sub-model to be spliced ​​is used as the training goal to obtain the environmental model, including: based on the first loss function corresponding to the water pump flow prediction sub-model to be spliced, the predicted water pump flow and the current training sample are subjected to loss processing to obtain a first loss value, and based on the first loss value, the model parameters of the water pump flow prediction sub-model to be spliced ​​are corrected; based on the second loss function corresponding to the water pump flow prediction sub-model to be spliced, the predicted water pump pressure and the current training sample are subjected to loss processing to obtain a second loss value, and based on the second loss value, the water pump pressure prediction sub-model to be spliced ​​is subjected to loss processing The model parameters are corrected; based on the third loss function corresponding to the pipeline flow prediction submodel to be spliced, the predicted pipeline flow and the current training sample are subjected to loss processing to obtain a third loss value, and based on the third loss value, the model parameters of the pipeline flow prediction submodel to be spliced ​​are corrected; based on the fourth loss function corresponding to the pipeline pressure prediction submodel to be spliced, the predicted pipeline pressure and the current training sample are subjected to loss processing to obtain a fourth loss value, and based on the fourth loss value, the model parameters of the pipeline pressure prediction submodel to be spliced ​​are corrected; when it is detected that the first loss function, the second loss function, the third loss function, and the fourth loss function have all converged, the environmental model is obtained.

[0107] In this embodiment, each sub-model to be spliced ​​can be a generative adversarial network composed of a generator and a discriminator. Correspondingly, the loss function of each sub-model to be spliced ​​can include a generation loss function and a discriminant loss function.

[0108] In practical applications, after obtaining the predicted water pump flow, the first matrix to be processed can be constructed based on the sample water pump flow, sample water pump frequency, sample water well liquid level, sample pipeline pressure and water pump predicted flow in the current training sample, and the first matrix to be processed and the actual water pump flow are subjected to loss processing according to the first discriminant loss function in the first loss function to obtain the first discriminant loss value, and the model parameters of the discriminator in the water pump flow prediction sub-model to be spliced ​​are corrected based on the first discriminant loss value. Afterwards, the predicted water pump flow and the actual water pump flow in the current training sample can be subjected to loss processing according to the first generation loss function in the first loss function to obtain the first generation loss value, and the model parameters of the generator in the water pump flow prediction sub-model to be spliced ​​can be corrected based on the first generation loss value.

[0109] Furthermore, after obtaining the predicted water pump pressure, the predicted water pump pressure and the actual water pump pressure in the current training sample can be subjected to loss processing according to the second generation loss function in the second loss function to obtain a second generation loss value, and the model parameters of the generator in the water pump pressure prediction sub-model to be spliced ​​can be corrected based on the second generation loss value. Afterwards, a second matrix to be processed can be constructed based on the sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump frequency, predicted water pump flow and predicted water pump pressure in the current training sample, and the second matrix to be processed and the actual water pump pressure can be subjected to loss processing according to the second discrimination loss function in the second loss function to obtain a second discrimination loss value, and the model parameters of the discriminator in the water pump pressure prediction sub-model to be spliced ​​can be corrected based on the second discrimination loss value.

[0110] Furthermore, after obtaining the predicted pipeline flow, the predicted pipeline flow and the actual pipeline flow in the current training sample can be subjected to loss processing according to the third generation loss function in the third loss function to obtain a third generation loss value, and the model parameters of the generator in the pipeline flow prediction sub-model to be spliced ​​can be corrected based on the third generation loss value. Afterwards, a third matrix to be processed can be constructed based on the sample pipeline flow, sample water pump pressure, predicted water pump flow and predicted pipeline flow in the current training sample, and the third matrix to be processed and the actual pipeline flow can be subjected to loss processing according to the third discrimination loss function in the third loss function to obtain a third discrimination loss value, and the model parameters of the discriminator in the pipeline flow prediction sub-model to be spliced ​​can be corrected based on the third discrimination loss value.

[0111] Furthermore, after obtaining the predicted pipeline pressure, the predicted pipeline pressure and the actual pipeline pressure in the current training sample can be subjected to loss processing according to the fourth generation loss function in the fourth loss function to obtain a fourth generation loss value, and the model parameters of the generator in the pipeline pressure prediction sub-model to be spliced ​​can be corrected based on the fourth generation loss value. Afterwards, a fourth matrix to be processed can be constructed based on the sample pipeline flow, sample pipeline pressure, sample water pump pressure, predicted water pump pressure, predicted pipeline flow, and predicted pipeline pressure in the current training sample, and the fourth matrix to be processed and the actual pipeline pressure can be subjected to loss processing according to the fourth discrimination loss function in the fourth loss function to obtain a fourth discrimination loss value, and the model parameters of the discriminator in the pipeline pressure prediction sub-model to be spliced ​​can be corrected based on the fourth discrimination loss value.

[0112] Specifically, when using the loss value to correct the model parameters in each sub-model to be spliced, the convergence of the loss function can be used as a training goal, such as whether the training error is less than the preset error, or whether the error change tends to be stable, or whether the current number of iterations is equal to the preset number. If the detection reaches the convergence condition, such as the training error of the loss function is less than the preset error, or the error change trend tends to be stable, it indicates that the training of the sub-model to be spliced ​​is completed, and the iterative training can be stopped at this time. If it is detected that the convergence condition is not met at present, other training samples can be further obtained to continue training the sub-model to be spliced ​​until the training error of the loss function is within the preset range. When it is detected that the training errors of the loss functions in each sub-model to be spliced ​​have reached convergence, the trained model to be trained can be used as the environmental model, and an environmental model including a water pump flow prediction sub-model, a water pump pressure prediction sub-model, a pipeline flow prediction sub-model and a pipeline pressure prediction sub-model can be obtained.

[0113] Furthermore, the water pump pressure prediction sub-model in the environmental model can be used to predict the water pump pressure of the water pump to be tested in the pump room at different preset water pump flow rates at the next moment to obtain the predicted water pump pressure.

[0114] The technical solution of the embodiment of the present invention obtains multiple training samples, and then, for each training sample, inputs the current training sample into the model to be trained to obtain the output results corresponding to each sub-model to be spliced, and further, based on the output results, the model parameters in the corresponding sub-model to be spliced ​​are corrected, and the convergence of the loss function in each sub-model to be spliced ​​is used as the training goal to obtain the environmental model, and the water pump pressure prediction sub-model in the environmental model is used to predict the water pump pressure of the water pump to be tested in the pump room at the next moment under different preset water pump flow rates to obtain the predicted water pump pressure. By jointly training the sub-models to be spliced, the correlation between the environmental model and the actual operating conditions is improved, thereby improving the prediction accuracy of the environmental model for the pump room operation data.

[0115] Embodiment 4

[0116] Figure 5 Schematic diagram of the structure of a pump performance curve generating device provided by the fourth embodiment of the present invention. Figure 5 As shown, the device includes: a data acquisition module 410 , a preset data acquisition module 420 , a data processing module 430 and a performance curve determination module 440 .

[0117] Among them, the data acquisition module 410 is used to obtain the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested for each water pump to be tested in the pump room; wherein the pending operation data includes at least one of the water pump frequency, water pump pressure and water pump flow; the pending operation data includes the current operation data at the current moment and the historical operation data within a preset time period before the current moment; the pending liquid level data includes the current liquid level data at the current moment and the historical liquid level data within the preset time period before the current moment; the preset data acquisition module 420 is used to obtain the next moment of the current water pump to be tested at the current moment The preset water pump frequency and preset water pump flow interval; a data processing module 430 is used to process the to-be-processed operating data, the to-be-processed liquid level data, the preset water pump frequency and the preset water pump flow interval based on the pre-retrieved water pump pressure prediction sub-model, and obtain a predicted water pump pressure interval corresponding to the preset water pump flow interval; wherein the water pump pressure prediction sub-model is a sub-model in the environmental model, the environmental model is constructed by at least one sub-model, and the at least one sub-model is spliced ​​in a preset splicing order; the performance curve determination module 440 determines the pump performance curve of the current water pump to be tested at the next moment based on the preset water pump flow interval and the predicted water pump pressure interval.

[0118] The technical solution of the embodiment of the present invention obtains the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested for each water pump to be tested in the pump room; then, obtains the preset water pump frequency and preset water pump flow range of the current water pump to be tested at the next moment of the current moment; further, based on the pre-retrieved water pump pressure prediction sub-model, the pending operation data, the pending liquid level data, the preset water pump frequency and the preset water pump flow range are processed to obtain the predicted water pump pressure range corresponding to the preset water pump flow range; finally, based on the preset water pump flow range and the predicted water pump pressure range, the pump performance curve of the current water pump to be tested at the next moment is determined, which solves the problems of increasing energy consumption and maintenance costs when generating pump performance curves in related technologies, failing to accurately predict the performance of the water pump in actual working conditions, and failing to cover all possible working conditions, and realizes the effect of simulating and generating pump room operation data at the predicted moment based on the historical operation data of the water pump, increases the diversity and richness of the water pump performance analysis data, thereby improving the accuracy of the pump performance curve, and realizing the effect of real-time reflection of the dynamic changes of the pump room system.

[0119] Optionally, the data processing module 430 includes: an interval division unit, a data processing unit and a prediction interval determination unit.

[0120] an interval division unit, configured to divide the preset water pump flow interval based on a pre-division step length to obtain a plurality of preset water pump flow rates, and construct a set of data to be processed based on the operation data to be processed, the liquid level data to be processed, the preset water pump frequency and the preset water pump flow rate to obtain a plurality of sets of data to be processed;

[0121] A data processing unit, configured to process the plurality of groups of data to be processed based on the water pump pressure prediction sub-model to obtain a water pump predicted pressure corresponding to each group of data to be processed;

[0122] A prediction interval determination unit is used to determine the predicted water pump pressure interval based on the water pump predicted pressures corresponding to the multiple groups of data to be processed.

[0123] Optionally, the performance curve determination module 440 includes: a lift interval determination unit and a performance curve generation unit.

[0124] A head interval determination unit, used to determine a predicted water pump head interval corresponding to the predicted water pump pressure interval;

[0125] The performance curve generating unit is used to generate a pump performance curve of the current water pump to be tested at the next moment based on the preset water pump flow interval and the predicted water pump head interval.

[0126] The pump performance curve generating device provided in the embodiment of the present invention can execute the pump performance curve generating method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0127] Embodiment 5

[0128] Figure 6 Schematic diagram of a model building device provided in Embodiment 5 of the present invention. Figure 6 As shown, the device includes: a sub-model acquisition module 510 and a sub-model splicing module 520.

[0129] Among them, the sub-model acquisition module 510 is used to obtain multiple sub-models to be spliced; the sub-model splicing module 520 is used to splice the sub-models to be spliced ​​together according to a preset splicing order to obtain a model to be trained.

[0130] The technical solution of the embodiment of the present invention obtains multiple sub-models to be spliced, and further splices the sub-models to be spliced ​​together according to a preset splicing order to obtain a model to be trained, thereby achieving the effect of constructing an environmental model to be trained corresponding to the operating conditions of the pump room. Furthermore, it can achieve the effect of real-time tracking of the pump room operating data and real-time export of the performance curve of the water pump based on the trained environmental model.

[0131] Optionally, the multiple sub-models to be spliced ​​include a water pump flow prediction sub-model to be spliced, a water pump pressure prediction sub-model to be spliced, a pipeline flow prediction sub-model to be spliced, and a pipeline pressure prediction sub-model to be spliced; the sub-model splicing module 520 is specifically used to use the model output of the water pump flow prediction sub-model to be spliced ​​as the model input of the water pump pressure prediction sub-model to be spliced ​​and the pipeline flow prediction sub-model to be spliced; and use the model output of the water pump pressure prediction sub-model to be spliced ​​and the model output of the pipeline flow prediction sub-model to be spliced ​​as the model input of the pipeline pressure prediction sub-model to be spliced.

[0132] The model building device provided in the embodiment of the present invention can execute the model building method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0133] Embodiment 6

[0134] Figure 7 Schematic diagram of the structure of a model training device provided by Embodiment 6 of the present invention. Figure 7 As shown, the device includes: a training sample acquisition module 610, an output result determination module 620 and a model parameter correction module 630.

[0135] Among them, the training sample acquisition module 610 is used to obtain multiple training samples; wherein the training samples include sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump flow and actual water pump pressure corresponding to the prediction time, sample pipeline flow, sample pipeline pressure, actual pipeline flow and actual pipeline pressure corresponding to the prediction time and sample water absorption well liquid level; the output result determination module 620 is used to input the current training sample into the model to be trained for each training sample, and obtain the output result corresponding to each sub-model to be spliced; the model parameter correction module 630 is used to correct the model parameters in the corresponding sub-model to be spliced ​​based on the output result, and take the convergence of the loss function in each sub-model to be spliced ​​as the training target to obtain the environmental model, so as to predict the water pump pressure of the water pump to be tested in the pump room at different preset water pump flows at the next moment based on the water pump pressure prediction sub-model in the environmental model, and obtain the predicted water pump pressure.

[0136] The technical solution of the embodiment of the present invention obtains multiple training samples, and then, for each training sample, inputs the current training sample into the model to be trained to obtain the output results corresponding to each sub-model to be spliced, and further, based on the output results, the model parameters in the corresponding sub-model to be spliced ​​are corrected, and the convergence of the loss function in each sub-model to be spliced ​​is used as the training goal to obtain the environmental model, and the water pump pressure prediction sub-model in the environmental model is used to predict the water pump pressure of the water pump to be tested in the pump room at the next moment under different preset water pump flow rates to obtain the predicted water pump pressure. By jointly training the sub-models to be spliced, the correlation between the environmental model and the actual operating conditions is improved, thereby improving the prediction accuracy of the environmental model for the pump room operation data.

[0137] Optionally, the output results include predicted water pump flow, predicted water pump pressure, predicted pipeline flow and predicted pipeline pressure; the output result determination module 620 includes: a predicted water pump flow determination unit, a predicted water pump pressure determination unit, a predicted pipeline flow determination unit and a predicted pipeline pressure determination unit.

[0138] A predicted water pump flow determination unit, used for inputting the sample water pump flow, sample water pump frequency, sample water suction well liquid level and sample pipeline pressure in the current training sample into the to-be-joined water pump flow prediction sub-model in the to-be-trained model, to obtain the predicted water pump flow corresponding to the current training sample;

[0139] A predicted water pump pressure determination unit, used for inputting the sample water pump flow, the sample water pump pressure, the sample water pump frequency, the actual water pump frequency and the predicted water pump flow into the water pump pressure prediction sub-model to be spliced ​​in the model to be trained, to obtain the predicted water pump pressure corresponding to the current training sample;

[0140] A predicted pipeline flow determination unit, used for inputting the sample pipeline flow, the sample water pump pressure and the predicted water pump flow into the pipeline flow prediction sub-model to be spliced ​​in the model to be trained, to obtain the predicted pipeline flow corresponding to the current training sample;

[0141] The predicted pipeline pressure determination unit is used to input the sample pipeline flow, sample pipeline pressure, sample water pump pressure, the predicted water pump pressure and the predicted pipeline flow into the pipeline pressure prediction sub-model to be spliced ​​in the model to be trained to obtain the predicted pipeline pressure corresponding to the current training sample.

[0142] Optionally, the model parameter correction module 630 includes: a first loss value determination unit, a second loss value determination unit, a third loss value determination unit, a fourth loss value determination unit and an environment model determination unit.

[0143] A first loss value determining unit, configured to perform loss processing on the predicted water pump flow and the current training sample based on a first loss function corresponding to the water pump flow prediction sub-model to be spliced, obtain a first loss value, and modify model parameters of the water pump flow prediction sub-model to be spliced ​​based on the first loss value;

[0144] a second loss value determining unit, configured to perform loss processing on the predicted water pump pressure and the current training sample based on a second loss function corresponding to the water pump flow prediction sub-model to be spliced, to obtain a second loss value, and to modify model parameters of the water pump pressure prediction sub-model to be spliced ​​based on the second loss value;

[0145] a third loss value determining unit, configured to perform loss processing on the predicted pipeline flow and the current training sample based on a third loss function corresponding to the pipeline flow prediction sub-model to be spliced, obtain a third loss value, and modify model parameters of the pipeline flow prediction sub-model to be spliced ​​based on the third loss value;

[0146] a fourth loss value determining unit, configured to perform loss processing on the predicted pipeline pressure and the current training sample based on a fourth loss function corresponding to the pipeline pressure prediction sub-model to be spliced, obtain a fourth loss value, and modify model parameters of the pipeline pressure prediction sub-model to be spliced ​​based on the fourth loss value;

[0147] The environment model determination unit is used to obtain the environment model when it is detected that the first loss function, the second loss function, the third loss function, and the fourth loss function have all converged.

[0148] The model training device provided in the embodiment of the present invention can execute the model training method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0149] Embodiment 7

[0150] Figure 8 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0151] like Figure 8 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0152] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0153] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as a pump performance curve generation method, a model building, and a training method.

[0154] In some embodiments, the pump performance curve generation method, model building and training method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the pump performance curve generation method, model building and training method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the pump performance curve generation method, model building and training method in any other appropriate manner (for example, by means of firmware).

[0155] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0156] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0157] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0158] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0159] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0160] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0161] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0162] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating a pump performance curve, characterized in that: include: For each water pump to be tested in the pump room, obtain the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested; wherein the pending operation data include water pump frequency, water pump pressure and water pump flow; the pending operation data include current operation data at the current moment and historical operation data within a preset time period before the current moment; the pending liquid level data include current liquid level data at the current moment and historical liquid level data within the preset time period before the current moment; Obtaining a preset water pump frequency and a preset water pump flow range of the current water pump to be tested at a moment next to the current moment; Based on the pre-retrieved water pump pressure prediction sub-model, the to-be-processed operating data, the to-be-processed liquid level data, the preset water pump frequency and the preset water pump flow interval are processed to obtain a predicted water pump pressure interval corresponding to the preset water pump flow interval; wherein the water pump pressure prediction sub-model is a sub-model in the environmental model, the environmental model is constructed by at least one sub-model, and the at least one sub-model is spliced ​​according to a preset splicing order; Determining a pump performance curve of the current water pump to be tested at the next moment based on the preset water pump flow interval and the predicted water pump pressure interval; The method processes the operation data to be processed, the liquid level data to be processed, the preset water pump frequency and the preset water pump flow interval based on the pre-retrieved water pump pressure prediction sub-model to obtain a predicted water pump pressure interval corresponding to the preset water pump flow interval, including: Dividing the preset water pump flow interval based on a pre-divided step length to obtain a plurality of preset water pump flow rates, and constructing a set of data to be processed based on the operation data to be processed, the liquid level data to be processed, the preset water pump frequency and the preset water pump flow rate to obtain a plurality of sets of data to be processed; Processing the plurality of groups of data to be processed based on the water pump pressure prediction sub-model to obtain a water pump predicted pressure corresponding to each group of data to be processed; Determining the predicted water pump pressure range based on the predicted water pump pressures corresponding to the multiple groups of data to be processed; The environment model is obtained by training the model to be trained, and the method for constructing the model to be trained is: Get multiple sub-models to be spliced; Splicing the sub-models to be spliced ​​together according to a preset splicing order to obtain a model to be trained; The multiple sub-models to be spliced ​​include a water pump flow prediction sub-model to be spliced, a water pump pressure prediction sub-model to be spliced, a pipeline flow prediction sub-model to be spliced, and a pipeline pressure prediction sub-model to be spliced; each sub-model to be spliced ​​is spliced ​​together according to a preset splicing order to obtain a model to be trained, including: Using the model input of the model to be trained as the model input of the water pump flow prediction sub-model to be spliced, and using the model output of the water pump flow prediction sub-model to be spliced ​​as at least part of the model input of the water pump pressure prediction sub-model to be spliced ​​and the pipeline flow prediction sub-model to be spliced; The model output of the water pump pressure prediction sub-model to be spliced ​​and the model output of the pipeline flow prediction sub-model to be spliced ​​are used as at least part of the model input of the pipeline pressure prediction sub-model to be spliced, and the model output of each of the sub-models to be spliced ​​is used as the model output of the model to be trained, so as to obtain the model to be trained.

2. The method according to claim 1, characterized in that The step of determining the pump performance curve of the current water pump to be tested at the next moment based on the preset water pump flow interval and the predicted water pump pressure interval includes: Determining a predicted water pump head interval corresponding to the predicted water pump pressure interval; Based on the preset water pump flow interval and the predicted water pump head interval, a pump performance curve of the current water pump to be tested at the next moment is generated.

3. The method according to claim 1, characterized in that The model training method of the model to be trained includes: Acquire multiple training samples; wherein the training samples include sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump flow and actual water pump pressure corresponding to the prediction time, sample pipeline flow, sample pipeline pressure, actual pipeline flow and actual pipeline pressure corresponding to the prediction time, and sample water absorption well liquid level; For each training sample, the current training sample is input into the model to be trained to obtain the output results corresponding to each sub-model to be spliced; Based on the output result, the model parameters in the corresponding sub-models to be spliced ​​are corrected, and the convergence of the loss function in each of the sub-models to be spliced ​​is used as a training target to obtain an environmental model, so as to predict the water pump pressure of the water pump to be tested in the pump room at different preset water pump flow rates at the next moment based on the water pump pressure prediction sub-model in the environmental model to obtain a predicted water pump pressure; The output results include predicted water pump flow, predicted water pump pressure, predicted pipeline flow and predicted pipeline pressure; the current training sample is input into the model to be trained to obtain the actual output results corresponding to each sub-model to be spliced, including: Input the sample water pump flow, sample water pump frequency, sample water well liquid level and sample pipeline pressure in the current training sample into the water pump flow prediction sub-model to be spliced ​​in the model to be trained, and obtain the predicted water pump flow corresponding to the current training sample; Input the sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump frequency and predicted water pump flow into the water pump pressure prediction sub-model to be spliced ​​in the model to be trained, and obtain the predicted water pump pressure corresponding to the current training sample; Input the sample pipeline flow, sample water pump pressure and the predicted water pump flow into the pipeline flow prediction sub-model to be spliced ​​in the model to be trained to obtain the predicted pipeline flow corresponding to the current training sample; The sample pipeline flow, sample pipeline pressure, sample water pump pressure, the predicted water pump pressure and the predicted pipeline flow are input into the pipeline pressure prediction sub-model to be spliced ​​in the model to be trained to obtain the pipeline prediction pressure corresponding to the current training sample.

4. The method according to claim 3, characterized in that Based on the actual output results, the model parameters in the corresponding sub-models to be spliced ​​are modified, and the convergence of the loss functions in each of the sub-models to be spliced ​​is used as a training target to obtain an environment model, including: Based on the first loss function corresponding to the water pump flow prediction sub-model to be spliced, the predicted water pump flow and the current training sample are subjected to loss processing to obtain a first loss value, and based on the first loss value, the model parameters of the water pump flow prediction sub-model to be spliced ​​are corrected; Based on the second loss function corresponding to the water pump flow prediction sub-model to be spliced, the predicted water pump pressure and the current training sample are subjected to loss processing to obtain a second loss value, and based on the second loss value, the model parameters of the water pump pressure prediction sub-model to be spliced ​​are corrected; Based on the third loss function corresponding to the pipeline flow prediction sub-model to be spliced, the predicted pipeline flow and the current training sample are subjected to loss processing to obtain a third loss value, and based on the third loss value, the model parameters of the pipeline flow prediction sub-model to be spliced ​​are corrected; Performing loss processing on the predicted pipeline pressure and the current training sample based on a fourth loss function corresponding to the pipeline pressure prediction sub-model to be spliced ​​to obtain a fourth loss value, and modifying model parameters of the pipeline pressure prediction sub-model to be spliced ​​based on the fourth loss value; When it is detected that the first loss function, the second loss function, the third loss function, and the fourth loss function all converge, the environment model is obtained.

5. A pump performance curve generating device, characterized in that: The device comprises: A data acquisition module is used to obtain, for each water pump to be tested in the pump room, the pending operation data corresponding to the current water pump to be tested and the pending liquid level data corresponding to the target water suction well associated with the current water pump to be tested; wherein the pending operation data include water pump frequency, water pump pressure and water pump flow; the pending operation data include current operation data at the current moment and historical operation data within a preset time period before the current moment; the pending liquid level data include current liquid level data at the current moment and historical liquid level data within the preset time period before the current moment; A preset data acquisition module, used to acquire a preset water pump frequency and a preset water pump flow range of the current water pump to be tested at a moment next to the current moment; A data processing module, used for processing the to-be-processed operating data, the to-be-processed liquid level data, the preset water pump frequency and the preset water pump flow interval based on a pre-retrieved water pump pressure prediction sub-model, to obtain a predicted water pump pressure interval corresponding to the preset water pump flow interval; wherein the water pump pressure prediction sub-model is a sub-model in the environmental model, the environmental model is constructed by at least one sub-model, and the at least one sub-model is spliced ​​according to a preset splicing order; A performance curve determination module, used to determine the pump performance curve of the current water pump to be tested at the next moment based on the preset water pump flow interval and the predicted water pump pressure interval; The data processing module includes: an interval division unit, a data processing unit and a prediction interval determination unit; The interval division unit is used to divide the preset water pump flow interval based on a pre-division step length to obtain multiple preset water pump flow rates, and construct a group of data to be processed based on the operation data to be processed, the liquid level data to be processed, the preset water pump frequency and the preset water pump flow rate to obtain multiple groups of data to be processed; The data processing unit is used to process the multiple groups of data to be processed based on the water pump pressure prediction sub-model to obtain the water pump predicted pressure corresponding to each group of data to be processed; The prediction interval determination unit is used to determine the predicted water pump pressure interval based on the water pump predicted pressures corresponding to the multiple groups of data to be processed; The environment model is obtained by training the model to be trained, and the method for constructing the model to be trained includes: a sub-model acquisition module, which is used to obtain multiple sub-models to be spliced; A sub-model splicing module is used to splice the sub-models to be spliced ​​together according to a preset splicing order to obtain a model to be trained; The multiple sub-models to be spliced ​​include a water pump flow prediction sub-model to be spliced, a water pump pressure prediction sub-model to be spliced, a pipeline flow prediction sub-model to be spliced, and a pipeline pressure prediction sub-model to be spliced; the sub-model splicing module is specifically used to use the model input of the model to be trained as the model input of the water pump flow prediction sub-model to be spliced, and use the model output of the water pump flow prediction sub-model to be spliced ​​as at least part of the model input of the water pump pressure prediction sub-model to be spliced ​​and the pipeline flow prediction sub-model to be spliced; use the model output of the water pump pressure prediction sub-model to be spliced ​​and the model output of the pipeline flow prediction sub-model to be spliced ​​as at least part of the model input of the pipeline pressure prediction sub-model to be spliced, and use the model output of each of the sub-models to be spliced ​​as the model output of the model to be trained, so as to obtain the model to be trained.

6. The device according to claim 5, characterized in that The model training device of the model to be trained comprises: A training sample acquisition module, used to acquire multiple training samples; wherein the training samples include sample water pump flow, sample water pump pressure, sample water pump frequency, actual water pump flow and actual water pump pressure corresponding to the prediction time, sample pipeline flow, sample pipeline pressure, actual pipeline flow and actual pipeline pressure corresponding to the prediction time, and sample water absorption well liquid level; The output result determination module is used to input the current training sample into the model to be trained for each training sample, and obtain the output result corresponding to each sub-model to be spliced; A model parameter correction module is used to correct the model parameters in the corresponding sub-models to be spliced ​​based on the output results, and take the convergence of the loss functions in each of the sub-models to be spliced ​​as a training target to obtain an environmental model, so as to predict the water pump pressure of the water pump to be tested in the pump room at different preset water pump flow rates at the next moment based on the water pump pressure prediction sub-model in the environmental model to obtain a predicted water pump pressure; The output result includes a predicted water pump flow, a predicted water pump pressure, a predicted pipeline flow, and a predicted pipeline pressure; the output result determination module includes: a predicted water pump flow determination unit, a predicted water pump pressure determination unit, a predicted pipeline flow determination unit, and a predicted pipeline pressure determination unit; The predicted water pump flow determination unit is used to input the sample water pump flow, sample water pump frequency, sample water suction well liquid level and sample pipeline pressure in the current training sample into the to-be-joined water pump flow prediction sub-model in the to-be-trained model to obtain the predicted water pump flow corresponding to the current training sample; The predicted water pump pressure determination unit is used to input the sample water pump flow, the sample water pump pressure, the sample water pump frequency, the actual water pump frequency and the predicted water pump flow into the water pump pressure prediction sub-model to be spliced ​​in the model to be trained, to obtain the predicted water pump pressure corresponding to the current training sample; The predicted pipeline flow determination unit is used to input the sample pipeline flow, the sample water pump pressure and the predicted water pump flow into the pipeline flow prediction sub-model to be spliced ​​in the model to be trained, to obtain the predicted pipeline flow corresponding to the current training sample; The predicted pipeline pressure determination unit is used to input the sample pipeline flow, sample pipeline pressure, sample water pump pressure, the predicted water pump pressure and the predicted pipeline flow into the pipeline pressure prediction sub-model to be spliced ​​in the model to be trained to obtain the predicted pipeline pressure corresponding to the current training sample.

Images