A method and related apparatus for obtaining pump start-up parameters

Abstract

This invention discloses a method and related apparatus for obtaining pump start-up parameters. A machine learning model for each candidate water injection pump is pre-trained based on historical data of the water injection system. Multiple simulation calculations are executed. If the simulation calculation process meets the preset optimal energy consumption conditions, the candidate pump start-up combinations and the candidate frequency set corresponding to the simulation calculation process are used as the pump start-up parameters. Since the simulation calculation process obtains the total energy consumption of the candidate pump start-up combinations based on the demand parameters, the single pump frequency of each candidate water injection pump, and the machine learning model when the on / off state of each candidate water injection pump is "on," this is used as the calculation result of the simulation calculation process. Furthermore, since the optimal energy consumption condition includes the minimum calculation result among all simulation calculation results, this solution obtains pump start-up parameters that can reduce total energy consumption, which is beneficial for improving the operating efficiency of the water injection system and reducing its total energy consumption.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and related apparatus for obtaining pump start-up parameters. Background Technology

[0002] Injecting water into the oil reservoir using a water injection system to replenish and maintain reservoir pressure is a necessary means to achieve high and stable oil production and improve oil recovery. Typically, a water injection system consists of multiple water injection pumps operating in parallel. The total energy consumption of these pumps is a crucial factor in measuring water injection efficiency. Reducing the total energy consumption of these pumps and improving their operating efficiency are pressing issues that need to be addressed in this field. Summary of the Invention

[0003] In view of the above problems, the present invention provides a method and related apparatus for obtaining pump start-up parameters to overcome or at least partially solve the above problems.

[0004] A method for obtaining pump start-up parameters, comprising:

[0005] Obtain candidate pump start combinations and preset demand parameters. The candidate pump start combinations include at least one candidate water injection pump. The demand parameters include the water tank level, the external pressure, and the external flow rate. The candidate water injection pump is the water injection pump in the water injection system.

[0006] A machine learning model for each candidate water injection pump is pre-trained based on historical data of the water injection system, wherein the historical data includes historical demand parameters and historical single pump frequency.

[0007] The simulation calculation process is executed multiple times. Each simulation calculation process includes: obtaining a candidate frequency set, which includes the single pump frequency of each candidate water injection pump; when the switch state of each candidate water injection pump is on, obtaining the total energy consumption of the candidate pump-on combination based on the demand parameters, the single pump frequency of each candidate water injection pump and the machine learning model, as the calculation result of the simulation calculation process.

[0008] If the simulation calculation process meets the preset optimal energy consumption conditions, the candidate pump start-up combination and the candidate frequency set corresponding to the simulation calculation process are used as pump start-up parameters. The optimal energy consumption conditions include the minimum calculation result among all simulation calculation process calculation results.

[0009] Optionally, the machine learning model of the target water injection pump includes a first model, a second model, a third model, and a fourth model, wherein the target water injection pump is any candidate water injection pump, and the first model is a linear model;

[0010] The first model of the target water injection pump is pre-trained using the historical single pump frequency of the target water injection pump as the training input sample and the single pump flow rate of the target water injection pump when it is turned on at the historical single pump frequency as the training target.

[0011] The second model of the target water injection pump is pre-trained using historical water tank level and historical single pump flow rate as training input samples and the pump inlet pressure of the target water injection pump when it is running under the historical water tank level and historical single pump flow rate as the training target.

[0012] The third model of the target water injection pump is pre-trained using historical external pressure and historical single pump flow rate as training input samples, and the pump post-pump pressure of the target water injection pump when it is running under the historical water storage tank level and historical single pump flow rate as the training target.

[0013] The fourth model of the target water injection pump is pre-trained using the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as training input samples, and the energy consumption of the target water injection pump when operating under the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as the training target.

[0014] Optionally, a candidate frequency set is obtained, including:

[0015] Obtain the single pump frequency of the first type of candidate water injection pump in the candidate pump start combination, where the number of the first type of candidate water injection pump is n-1, and n is equal to the number of candidate water injection pumps in the candidate pump start combination;

[0016] Based on the single pump frequency and the first model of each of the first type of candidate water injection pumps, the single pump flow rate of each of the first type of candidate water injection pumps is obtained.

[0017] The difference between the outflow rate and the single pump flow rate of all the first type of candidate water injection pumps is taken as the single pump flow rate of the second type of candidate water injection pump. The second type of candidate water injection pump is the candidate water injection pump other than the first type of candidate water injection pump in the candidate pump start combination.

[0018] Based on the single-pump flow rate of the second type of candidate water injection pumps and the second model, the single-pump frequency of the second type of candidate water injection pumps is obtained.

[0019] Obtain the candidate frequency set, which includes the single pump frequency of each of the first type of candidate water injection pumps and the single pump frequency of each of the second type of candidate water injection pumps.

[0020] Optionally, the simulation calculation process can be executed multiple times, including: iteratively executing the simulation calculation process multiple times;

[0021] The optimal energy consumption conditions also include: the difference between the calculation results of a series of simulation calculations is less than a preset difference threshold, and / or the number of iterations is greater than a preset number threshold.

[0022] Optionally, the method further includes:

[0023] If the simulation calculation process does not meet the optimal energy consumption condition, the minimum calculation result is used as the optimization condition, and the single pump frequency of each first type of candidate water injection pump is updated using a preset optimization algorithm to obtain the update frequency of each first type of candidate water injection pump.

[0024] The step of obtaining the single pump frequency of the first type of candidate water injection pump in the candidate pump start-up combination includes:

[0025] If the simulation calculation process is the first simulation calculation process, the preset initial frequency of each of the first type of candidate water injection pumps is taken as the single pump frequency.

[0026] If the simulation calculation process is the kth iteration, the update frequency of each first-type candidate water injection pump obtained from the (k-1)th simulation calculation process is taken as the single pump frequency, where k is an integer greater than 1.

[0027] Optionally, based on the demand parameters and the machine learning model of each candidate water injection pump, the total energy consumption of the candidate pump-starting combination is obtained, including:

[0028] The water level in the storage tank and the single pump flow rate of each candidate water injection pump are correspondingly input into the second model of each candidate water injection pump, and the output of the second model of each candidate water injection pump is used as the pump inlet pressure of each candidate water injection pump.

[0029] The external pressure and the single pump flow rate of each candidate water injection pump are input into the third model of each candidate water injection pump, and the output of the third model of each candidate water injection pump is used as the pump outlet pressure of each candidate water injection pump.

[0030] The single pump flow rate, single pump frequency, pump inlet pressure and pump outlet pressure of each candidate water injection pump are input into the fourth model of each candidate water injection pump, and the output of the fourth model of each candidate water injection pump is used as the energy consumption of each candidate water injection pump.

[0031] The sum of the energy consumption of each of the candidate water injection pumps is taken as the total energy consumption.

[0032] Optionally, obtaining candidate pump start combinations includes: obtaining an optimization set, the optimization set including multiple pump start combinations, each pump start combination including at least one water injection pump;

[0033] Pump start combinations are obtained from the optimization set in a traversal manner and used as candidate pump start combinations;

[0034] The method further includes:

[0035] If the simulation calculation result of the target candidate pump start combination that satisfies the optimal energy consumption condition is the smallest, the pump start parameters corresponding to the target candidate pump start combination are taken as the optimal pump start parameters.

[0036] A device for acquiring pump start-up parameters, comprising:

[0037] The parameter acquisition unit is used to acquire candidate pump start combinations and preset demand parameters. The candidate pump start combinations include at least one candidate water injection pump. The demand parameters include the water storage tank level, the external output pressure, and the external output flow rate. The candidate water injection pump is a water injection pump in the water injection system.

[0038] The model building unit is used to pre-train a machine learning model for each candidate water injection pump based on historical data of the water injection system. The historical data includes historical demand parameters and historical single pump frequency.

[0039] The simulation calculation unit is used to execute multiple simulation calculation processes. Each simulation calculation process includes: obtaining a candidate frequency set, which includes the single pump frequency of each candidate water injection pump; when the switching state of each candidate water injection pump is on, obtaining the total energy consumption of the candidate pump-on combination based on the demand parameters, the single pump frequency of each candidate water injection pump, and the machine learning model, as the calculation result of the simulation calculation process.

[0040] The result acquisition unit is used to take the candidate pump start combination and the candidate frequency set corresponding to the simulation calculation process as pump start parameters if the simulation calculation process meets the preset optimal energy consumption conditions. The optimal energy consumption conditions include the minimum calculation result among the calculation results of all simulation calculation processes.

[0041] A storage medium comprising a stored program, wherein, when the program is executed, the device containing the storage medium is controlled to perform the aforementioned method for obtaining pump start-up parameters.

[0042] An electronic device includes at least one processor, at least one memory connected to the processor, and a bus; wherein the processor and the memory communicate with each other via the bus; the processor is used to call program instructions in the memory to execute the above-described method for obtaining pump start-up parameters.

[0043] By employing the above technical solution, this invention provides a method and related apparatus for obtaining pump start-up parameters. The method acquires candidate pump start-up combinations and preset demand parameters. The candidate pump start-up combinations include at least one candidate water injection pump. The demand parameters include the water tank level, external pressure, and external flow rate. The candidate water injection pumps are water injection pumps in a water injection system, and these pumps are connected in parallel. A machine learning model for each candidate water injection pump is pre-trained based on historical data from the water injection system. The historical data includes historical demand parameters and historical single-pump frequencies. Multiple simulation calculations are performed. If the simulation calculations meet preset optimal energy consumption conditions, the candidate pump start-up combinations and the candidate frequency set corresponding to the simulation calculations are used as the pump start-up parameters. Since each simulation calculation process, when each candidate water injection pump is in the on / off state, obtains the total energy consumption of the candidate pump combination based on the demand parameters, the single pump frequency of each candidate water injection pump, and the machine learning model, and uses this as the calculation result of the simulation calculation process, the total power consumption of the candidate water injection pumps in the candidate pump combination is highly accurate and consistent with actual working conditions when the pumps in the simulation system are connected in parallel, under the conditions of the demand parameters and the single pump frequency of each candidate water injection pump. Furthermore, since the optimal energy consumption condition includes the minimum calculation result among all simulation calculation processes, the simulation calculation process corresponding to the candidate pump combination and candidate frequency set in the pump start parameters satisfies the minimum calculation result among multiple simulation calculation processes. That is, the total energy consumption generated by the candidate pump combination when it is started under the candidate frequency set and demand parameters is minimized. In summary, this scheme obtains pump start parameters that can reduce total energy consumption, which is beneficial to improving the operating efficiency of the water injection system and reducing the total energy consumption of the water injection system.

[0044] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0045] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0046] Figure 1 A schematic diagram showing the relationship between the operating parameters of a water injection system is presented.

[0047] Figure 2 A flowchart illustrating a method for obtaining pump start-up parameters according to an embodiment of this application is shown;

[0048] Figure 3 A flowchart illustrating another method for obtaining pump start-up parameters provided in an embodiment of this application is shown.

[0049] Figure 4 A flowchart illustrating another method for obtaining pump start-up parameters provided in an embodiment of this application is shown.

[0050] Figure 5 A schematic diagram of a device for obtaining pump start-up parameters provided in an embodiment of this application is shown;

[0051] Figure 6 A schematic diagram of a device for obtaining pump start-up parameters provided in an embodiment of this application is shown. Detailed Implementation

[0052] During their research, the inventors discovered that the operating parameters of the oilfield water injection system (hereinafter referred to as the water injection system) are complex, such as... Figure 1 As shown, the operating parameters of each water injection pump include, but are not limited to, its on / off status (e.g., ...). Figure 1 The pump switch shown, the pressure before the pump, the pressure after the pump, the pressure of the main line after the pump, the pressure of the main line, the flow rate of the main line, the leakage, the pump flow rate, and the liquid level, are as follows: Figure 1 As shown in the diagram, the arrows illustrate the complex coupling relationships between the parameters, making overall modeling extremely difficult.

[0053] Through theoretical analysis and objective data verification, the main contradictions were identified, secondary information was ignored, and the operating parameters were decoupled, dividing the main relationships into a flow loop (mass conservation) and a pressure loop (energy conservation). Based on this, the inventors determined the operating logic of the water injection system, and for each water injection pump, the functional relationships between various parameters included:

[0054] Relationship 1: Single pump flow rate = f(pump on / off, single pump frequency).

[0055] Relationship 2: Pump inlet pressure = f(water tank level, single pump flow rate).

[0056] Relationship 3: Pump post-pressure = f(external output pressure, single pump flow rate).

[0057] Relationship 4: Single pump energy consumption = f(single pump flow rate, pump inlet pressure, pump outlet pressure, single pump frequency).

[0058] Where f() represents the functional relationship, and for clarity, the flow rate of each water injection pump is called the single pump flow rate, the frequency is called the single pump frequency, and the pump switch indicates the switch status value of the water injection pump. When the water injection pump is in the on state, the pump switch value is 1, and when the water injection pump is in the off state, the pump switch value is 0.

[0059] It is understandable that when the pump's on / off setting is 0, the single pump flow rate is 0. When the pump's on / off setting is 1, there is a functional relationship between the water injection pump's flow rate and frequency, i.e., Relationship 1. Specifically, under some optional operating conditions, for example, when the water injection pump is a reciprocating plunger pump, its flow rate is approximately proportional to the frequency. Furthermore, there is a functional relationship between the pump inlet pressure, the water tank level, and the single pump flow rate (i.e., Relationship 2); a functional relationship between the pump outlet pressure, the external output pressure, and the single pump flow rate (i.e., Relationship 3); and a functional relationship between the single pump energy consumption, the single pump flow rate, the pump inlet pressure, the pump outlet pressure, and the single pump frequency (i.e., Relationship 4).

[0060] Therefore, the flow rate of each water injection pump can be controlled by adjusting its pump frequency. This, in turn, controls the energy consumption of each pump.

[0061] Based on the above research findings, given the required parameters of the water injection system (including external flow rate, external pressure, and water tank level), the total energy consumption of multiple water injection pumps connected in parallel can be controlled by controlling the on / off state and single pump frequency of the pumps connected in parallel.

[0062] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0063] The method for obtaining pump start-up parameters provided in this application embodiment is applicable to, but not limited to, a water injection system including multiple water injection pumps connected in parallel. Optionally, a specific application scenario of this application is: after obtaining the demand parameters of the water injection system, obtaining the pump start-up parameters that minimize the total energy consumption of the reciprocating plunger pumps in the water injection system. The pump start-up parameters include the pump start-up combination and the single pump frequency of each water injection pump in the pump start-up combination. The pump start-up combination includes each water injection pump configured to be in the on / off state during the operation of the water injection system. It should be noted that, in this embodiment, the pump start-up parameters are output as configuration parameters to indicate the configuration during the operation of the water injection system.

[0064] Taking a water injection system consisting of N parallel water pumps as an example, Figure 2 The specific implementation flow of a method for obtaining pump start-up parameters provided in this application embodiment is as follows: Figure 2 As shown, this method includes:

[0065] S201. Construct a machine learning model for each water injection pump based on historical data.

[0066] In this embodiment, the preset functional relationships include:

[0067] Relationship 1: Single pump flow rate = f(pump on / off, single pump frequency).

[0068] Relationship 2: Pump inlet pressure = f(water tank level, single pump flow rate).

[0069] Relationship 3: Pump post-pressure = f(external output pressure, single pump flow rate).

[0070] Relationship 4: Single pump energy consumption = f(single pump flow rate, pump inlet pressure, pump outlet pressure, single pump frequency).

[0071] Correspondingly, the machine learning model for each water pump includes a first model, a second model, a third model, and a fourth model.

[0072] The first model of the water injection pump is trained based on the first sample data and relation 1. The first sample data includes the first training input sample and the first training target. The first training input sample includes the sample switch state value and the sample single pump frequency. The first training target is the single pump flow rate of the water injection pump when it is running under the sample switch state value and the sample single pump frequency, which is called the sample single pump flow rate.

[0073] The second model of the water injection pump is trained based on the second sample data and relation 2. The second sample data includes the second training input sample and the second training target. The second training input sample includes the liquid level of the sample water tank and the flow rate of the sample single pump. The second training target is the pump inlet pressure when the water injection pump is running under the sample water tank liquid level and the sample single pump flow rate, which is called the sample pump inlet pressure.

[0074] The third model of the water injection pump is trained based on the third sample data and relation 3. The third sample data includes the third training input sample and the third training target. The third training input sample includes the sample external pressure and the sample single pump flow rate. The third training target is the pump post-pump pressure when the water injection pump is running under the sample external pressure and the sample single pump flow rate, which is called the sample pump post-pump pressure.

[0075] The fourth model of the water injection pump is trained based on the fourth sample data and relation 4. The fourth sample data includes the fourth training input samples and the fourth training objective. The fourth training input samples include the sample single pump flow rate, the sample pump inlet pressure, the sample pump outlet pressure, and the sample single pump frequency. The fourth training objective is the single pump energy consumption of the water injection pump when it is running under the sample single pump flow rate, sample pump inlet pressure, sample pump outlet pressure, and sample single pump frequency, which is called the sample single pump energy consumption.

[0076] It should be noted that the historical data includes the corresponding relationships of historical demand parameters, historical single pump frequency, historical single pump flow rate, historical inlet pressure, and historical outlet pressure. The first sample data, the second sample data, the third sample data, and the fourth sample data are all collected from historical data. The historical data are objective data obtained by monitoring the actual operating parameters of the water injection system.

[0077] Furthermore, it should be noted that machine learning models include, but are not limited to, linear models and neural network models. This application does not limit the specific structure of each model, and specific training methods can be found in the prior art.

[0078] Since the water injection pump in the specific application scenario of this embodiment is a reciprocating plunger pump, this embodiment takes the first model as a linear model and the second to fourth models as neural network models.

[0079] S202. Traverse the optimization set to obtain candidate pump-starting combinations.

[0080] In this embodiment, the optimization set includes at least one set of pump start combinations, and each pump start combination includes at least one water injection pump.

[0081] Specifically, the number of parallel water pumps is N, and the optimal set includes... This step involves obtaining all possible starting combinations of the N parallel water injection pumps.

[0082] In this embodiment, the pump start combinations in the optimization set are traversed. For each pump start combination, the optimization process is used to obtain the single pump frequency and minimum total energy consumption of each water injection pump that minimizes the total energy consumption when the switch state of the water injection pumps included in the pump start combination is on.

[0083] Taking a candidate pump-starting combination including water injection pumps 1 to n as an example, the preset on / off state of water injection pumps 1 to n is "on," while the on / off state of other water injection pumps (excluding water injection pumps 1 to n) is "off." The optimization process for the candidate pump-starting combination includes a multi-iteration simulation calculation process. It should be noted that the core of energy consumption optimization is to minimize energy consumption while ensuring that the water tank level, external flow rate, and external pressure are all within the required values. Therefore, the boundary conditions for optimization are the preset water tank level, preset external pressure, and preset external flow rate. The optimization parameters are the pump-starting combination and the single pump frequency of each water injection pump in the pump-starting combination. The optimization objective is to minimize the total energy consumption when the on / off state of n-1 water injection pumps in the pump-starting combination is "on."

[0084] Based on this, the simulation calculation process for any given simulation is shown in S203 to S213, as follows:

[0085] S203. Obtain the single pump frequency of n-1 water injection pumps in the candidate pump start combination.

[0086] In this embodiment, the method for obtaining the single pump frequency of n-1 water injection pumps in the candidate pump start-up combination includes:

[0087] In the first iteration, the method for obtaining the single pump frequencies of the n-1 injection pumps in the candidate pump-starting combinations is as follows: randomly generate a set of initial pump frequencies {f1, ... f2}. i ... f n-1}(1≤i≤n-1), and set the initial pump frequency f i The single pump frequency of water injection pump i.

[0088] In the kth (k>1)th iteration, the single pump frequency of the n-1 water injection pumps in the candidate pump-starting combination in the (k-1)th iteration is updated by the genetic algorithm to obtain the single pump frequency of the n-1 water injection pumps in the kth iteration.

[0089] It should be noted that the n-1 water injection pumps are randomly selected from the candidate pump start combinations. Optionally, the n-1 water injection pumps are denoted as the first type of water injection pumps, namely B1 to Bn-1.

[0090] S204. Based on the first model and single pump frequency of each type of water injection pump, obtain the single pump flow rate of each type of water injection pump.

[0091] In this embodiment, taking the first model M1 of the first type of water injection pump B1 as an example, the switch state value of B1 is equal to 1 and the single pump frequency of B1 is input into M1, and the output of M1 is obtained as the single pump flow rate of B1.

[0092] S205. Based on the preset external flow rate and the single pump flow rate of each first type of water injection pump, obtain the single pump flow rate of the second type of water injection pump.

[0093] In this embodiment, the second type of water injection pump is any water injection pump other than the first type in the candidate pump start-up combination. In actual production, the water injection system obtains the water injection flow rate and water injection pressure set by the downstream equipment, thereby obtaining the output flow rate and output pressure. Using the output flow rate and output pressure as demand parameters, water is injected into the downstream equipment. Since the water injection station supplies water to the downstream equipment through a single pipeline, the output flow rate is the total flow rate of all water injection pumps, which is also the total water injection flow rate of the water injection station for the downstream equipment.

[0094] Furthermore, based on the principle of flow conservation, the single-pump flow rate of the second type of water injection pump is equal to the difference between the external output flow rate and the single-pump flow rates of all first-type water injection pumps. Let Q be the single-pump flow rate of the second type of water injection pump. n Get Q n For specific methods, please refer to the following formula (1):

[0095]

[0096] Among them, Q i Bi represents the single-pump flow rate, and Q represents the external output flow rate.

[0097] S206. Based on the single-pump flow rate of the second type of water injection pump, obtain the single-pump frequency of the second type of water injection pump.

[0098] In this embodiment, based on the first model of the second type of water injection pump, the linear relationship between the single pump flow rate and the single pump frequency of the second type of water injection pump is obtained, and the single pump frequency of the second type of water injection pump is obtained based on the single pump flow rate and the linear relationship.

[0099] S207. Based on the preset water tank level, the single pump flow rate of each candidate water injection pump, and the second model, obtain the pump inlet pressure of each candidate water injection pump.

[0100] In this embodiment, the candidate water injection pump is the water injection pump in the candidate pump start combination, including the first type of water injection pump and the second type of water injection pump. Taking the second model M2 of the candidate water injection pump B1 as an example, the single pump flow rate of B1 and the water level of the storage tank are input to M2, and the output of M2 is obtained as the pump inlet pressure of B1.

[0101] S208. Based on the preset external pressure, the single pump flow rate of each candidate water injection pump, and the third model, obtain the pump post-pump pressure of each candidate water injection pump.

[0102] In this embodiment, taking the third model M3 of candidate water injection pump B1 as an example, the external pressure and the single pump flow rate of B1 are input to M3, and the output of M3 is obtained as the pump post-pressure of B1.

[0103] S209. Based on the single pump flow rate, single pump frequency, pump inlet pressure, and pump outlet pressure of each candidate water injection pump, obtain the energy consumption of each candidate water injection pump.

[0104] In this embodiment, taking the fourth model M4 of candidate water injection pump B1 as an example, the single pump flow rate, single pump frequency, pump inlet pressure, and pump outlet pressure of B1 are input to M4, and the output of M4 is obtained as the energy consumption of B1.

[0105] S210. The sum of the energy consumption of each candidate water injection pump is taken as the total energy consumption of the candidate pump start-up combination.

[0106] For example, let the energy consumption of candidate water injection pump Bj be W. j The total energy consumption W is calculated as follows:

[0107]

[0108] S211. Determine whether the preset stopping condition has been met.

[0109] In this embodiment, the stopping condition includes at least one of the following: the number of iterations reaches a preset threshold, or the difference between the total energy consumption obtained from the previous iteration and the total energy consumption is less than a preset difference threshold.

[0110] S212. If not, update the single pump frequency of the n-1 injection pumps in the candidate pump start combination and return to S203.

[0111] In this embodiment, the optimization objective is to minimize the total energy consumption of the candidate pump-starting combination. The single pump frequency of the n-1 water injection pumps in the candidate pump-starting combination is updated by the optimization algorithm. The on / off state of the n-1 water injection pumps is used as the start-up state, and the updated single pump frequency of the n-1 water injection pumps is used as the optimization parameters to execute the next iteration. The optimization algorithm includes a genetic algorithm, which can be found in the prior art.

[0112] S213. If yes, end the iteration and output the total energy consumption and the single pump frequency of each candidate water injection pump as the optimization result of the candidate pump start combination.

[0113] As described above, this embodiment traverses the optimization set and executes the iterative process of S203 to S213 for each pump start combination to obtain the optimization result for each pump start combination.

[0114] S214. Based on the optimization results of each pump start-up combination, obtain the pump start-up parameters of the water injection system.

[0115] Specifically, the total energy consumption in the optimization results of each pump start-up combination is compared to obtain the minimum total energy consumption. The optimization result including the minimum total energy consumption is taken as the optimal optimization result, and the single pump frequency and total energy consumption of each water injection pump included in the optimal optimization result are taken as the pump start-up parameters of the water injection system.

[0116] As can be seen from the above technical solutions, the method for obtaining pump start-up parameters provided in this application embodiment can achieve at least the following beneficial effects:

[0117] First, iterate through all possible pump-starting combinations and use an iterative optimization algorithm to obtain the optimization result for each pump-starting combination. Based on the optimization results of each pump-starting combination, obtain the pump-starting parameters of the water injection system. These pump-starting parameters are used to set the on / off state of the water injection pumps in the water injection system and control the frequency of the pumps that are turned on. Therefore, by obtaining the pump-starting parameters that minimize the total energy consumption from the possible pump-starting combinations, and outputting these pump-starting parameters as configuration parameters, the operating state of the water injection system can be controlled based on these parameters, thereby reducing the energy consumption of the water injection system and improving the water injection efficiency.

[0118] Second, based on historical data, each model in the machine learning model of each water injection pump is constructed. By learning from historical data through machine learning algorithms, the functional relationship between the parameters of the water injection pump under the current structure of the water injection system is found, resulting in a regression model (i.e., a machine learning model) used to calculate each target parameter. This replaces the empirical formulas in the existing technology, improves the accuracy of the calculation of the target parameters of each water injection pump (including single pump flow, energy consumption, pump downstream pressure, and pump upstream pressure), and thus improves the accuracy of optimization.

[0119] Third, during the iteration process, the single-pump frequencies of n-1 water injection pumps (the first type of water injection pumps) are used as the optimization space. The single-pump frequencies of the second type of water injection pumps are calculated by inversely deducing from the flow conservation principle. This avoids the large number of invalid samples that do not meet the constraints caused by adding all the single-pump frequencies of n water injection pumps to the optimization space, thus improving the optimization efficiency. For example, in a genetic algorithm, if all the single-pump frequencies of n water injection pumps are added to the optimization space, a larger initial population and more iterations are needed to generate a sufficient effective population, resulting in very low optimization efficiency. This method can generate a large number of samples that meet the constraints without requiring a large initial population or many iterations, greatly improving the optimization efficiency.

[0120] Figure 3 The implementation flow of another method for obtaining pump start-up parameters provided in the embodiments of this application is as follows: Figure 3 As shown, the entire scheme can be divided into three steps: Step 1 model training, Step 2 simulation system construction, and Step 3 energy consumption optimization. Figure 4 The three steps described above are explained as follows:

[0121] Step 1: Use historical data to build a regression model through machine learning.

[0122] In this embodiment, the specific method of this step can be found in the above embodiment, and will not be repeated here.

[0123] Step 2: Simulation System Setup

[0124] Specifically, a simulation system for the water injection system is jointly built based on the equipment connection relationships, physical mechanisms, and regression models of the water injection system.

[0125] The connection relationships include the number of parallel water injection pumps and the connection between the water injection pumps and the main external pipeline. The physical mechanism is mainly the conservation of mass (conservation of flow rate).

[0126] The simulation system is used to calculate the flow rate of a single pump based on the pump frequency using a regression model, the pressure before the pump based on the water tank level and the pump flow rate, the pressure after the pump based on the output pressure and the pump flow rate, and the energy consumption of a single water injection pump based on the pump flow rate, the pressure before the pump, the pressure after the pump, and the pump frequency. Under given operating conditions (output pressure, water tank level, output flow rate), the system calculates the flow rate, pressure, energy consumption, and other indicators of the water injection system.

[0127] The specific functional implementation of the simulation system can be found in the above embodiments.

[0128] Step 3: Using a genetic algorithm, under the simulation system and operating conditions, the optimization objective is to minimize total energy consumption, and the recommended operating settings are obtained based on the optimization results.

[0129] The specific implementation method of this step is described in the above embodiments, and will not be repeated here.

[0130] As can be seen, this invention uses machine learning algorithms to support the construction of a high-precision simulation system, and then uses a simulation system to support the optimization process based on genetic algorithms, so as to more effectively describe the energy consumption under various combinations, thereby finding the pump start combination and pump start frequency with the minimum energy consumption under given water injection flow rate (external flow rate) and water injection pressure (external pressure).

[0131] It should be noted that, Figure 2 or Figure 3 This is merely one optional implementation process for obtaining pump start-up parameters provided in this application embodiment; this application also includes other specific implementation methods.

[0132] In some optional embodiments, the optimization algorithm is not limited to genetic algorithms, but can also be gradient descent algorithms and particle swarm optimization algorithms, etc.

[0133] In some optional embodiments, the structure of each model can include multiple types, not limited to the first model mentioned in the above embodiments being a linear model and other models being neural network models. For example, if all models are neural network models, then in the above embodiments, a step of obtaining a fifth model is added. The method for obtaining the fifth model is: training based on the fifth sample data and relation 5 (relation 5, single pump frequency = f(pump switching, single pump flow rate)). The fifth sample data includes the fifth training input samples and the fifth training target. The fifth training input samples include sample switching state values ​​and sample single pump flow rates. The first training target is the single pump flow rate of the water injection pump when it is running under the sample switching state values ​​and sample single pump flow rates, which is called the sample single pump frequency. Based on this, the implementation method of step S206, which obtains the single pump frequency of the second type of water injection pump based on the single pump flow rate of the second type of water injection pump, includes: inputting the single pump flow rate and switching state value (which is 1) of the second type of water injection pump into the fifth model, and obtaining the output of the fifth model as the single pump frequency of the second type of water injection pump.

[0134] In some optional embodiments, the training method of the model includes other optional implementations. For example, each training input sample may also include other sample parameters, which can be adjusted according to the working conditions.

[0135] In summary, the method for obtaining pump start-up parameters provided in the embodiments of this application can be summarized as follows: Figure 4 The process shown is as follows: Figure 4 As shown, this method includes:

[0136] S401, Obtain candidate pump start combinations and preset requirement parameters.

[0137] In this embodiment, the candidate pump start combination includes at least one candidate water injection pump. The required parameters include the water level in the storage tank, the external pressure, and the external flow rate. The candidate water injection pump is a water injection pump in the water injection system, and the water injection pumps in the water injection system are connected in parallel.

[0138] It should be noted that the method for obtaining candidate pump start combinations includes randomly selecting one group from a preset number of pump start combinations each time, or iteratively selecting candidate pump start combinations from a preset number of pump start combinations, or the method for setting the required parameters can be found in the prior art.

[0139] S402. A machine learning model for each candidate water injection pump is pre-trained based on historical data of the water injection system.

[0140] In this embodiment, historical data includes historical demand parameters and historical single-pump frequency.

[0141] Specifically, the machine learning model for the candidate water injection pump includes multiple models, and the training objective for any model of each candidate water injection pump is the historical operating parameters of the candidate water injection pump under the training input samples. See the above embodiment for specific model construction methods.

[0142] S403, Execute multiple simulation calculations.

[0143] In this embodiment, there are several methods for executing multiple simulation calculation processes. One optional method is to iteratively execute multiple simulation calculation processes. Each simulation calculation process includes:

[0144] 1. Obtain the candidate frequency set.

[0145] The candidate frequency set includes the single-pump frequency of each candidate water injection pump. Various methods exist for obtaining the single-pump frequency of each candidate water injection pump. Optionally, a preset initial frequency of each candidate water injection pump can be obtained as the single-pump frequency for the first simulation calculation process, and the updated single-pump frequency can be obtained based on the calculation results of the (k-1)th simulation calculation process to obtain the single-pump frequency for the kth simulation calculation process. See the above embodiments for details.

[0146] 2. When the switch state of each candidate water injection pump is on, the total energy consumption of the candidate pump combination is obtained based on the demand parameters, the single pump frequency of each candidate water injection pump and the machine learning model, and is used as the calculation result of the simulation calculation process.

[0147] S404. If the simulation calculation process meets the preset optimal energy consumption conditions, the candidate pump start-up combination and the candidate frequency set corresponding to the simulation calculation process shall be used as the pump start-up parameters.

[0148] In this embodiment, the optimal energy consumption condition includes the minimum calculation result across all simulation calculation processes.

[0149] As can be seen from the above technical solutions, the method for obtaining pump start-up parameters provided in this application, since each simulation calculation process, when the switching state of each candidate water injection pump is on, obtains the total energy consumption of the candidate pump start-up combination based on the demand parameters, the single pump frequency of each candidate water injection pump, and the machine learning model, as the calculation result of the simulation calculation process. The machine learning model of each candidate water injection pump is pre-trained based on historical data of the water injection system. Therefore, under the simulated operating conditions of parallel water injection pumps in the water injection system (including but not limited to connection methods and models), the accuracy of the total power consumption of the candidate water injection pumps in the candidate pump start-up combination under the conditions of the demand parameters and the single pump frequency of each candidate water injection pump is high, conforming to actual operating conditions. Furthermore, since the simulation calculation process corresponding to the candidate pump start-up combination and the candidate frequency set in the pump start-up parameters satisfies the requirement that the calculation result is minimized in multiple simulation calculation processes, that is, the total energy consumption generated by the candidate pump start-up combination under the candidate frequency set and demand parameters is minimized, this solution obtains pump start-up parameters that can reduce total energy consumption, which is beneficial to improving the operating efficiency of the water injection system and reducing the total energy consumption of the water injection system.

[0150] Figure 5 A schematic diagram of the structure of a pump start-up parameter acquisition device provided in this application embodiment is shown below. Figure 5 As shown, this device includes:

[0151] The parameter acquisition unit 501 is used to acquire candidate pump start combinations and preset demand parameters. The candidate pump start combinations include at least one candidate water injection pump. The demand parameters include water tank level, external pressure, and external flow rate. The candidate water injection pump is a water injection pump in the water injection system.

[0152] The model building unit 502 is used to pre-train a machine learning model for each candidate water injection pump based on the historical data of the water injection system. The historical data includes historical demand parameters and historical single pump frequency.

[0153] The simulation calculation unit 503 is used to execute multiple simulation calculation processes. Each simulation calculation process includes: obtaining a candidate frequency set, which includes the single pump frequency of each candidate water injection pump; when the switching state of each candidate water injection pump is on, obtaining the total energy consumption of the candidate pump-on combination based on the demand parameters, the single pump frequency of each candidate water injection pump, and the machine learning model, as the calculation result of the simulation calculation process.

[0154] The result acquisition unit 504 is used to take the candidate pump start combination and the candidate frequency set corresponding to the simulation calculation process as pump start parameters if the simulation calculation process meets the preset optimal energy consumption conditions. The optimal energy consumption conditions include the minimum calculation result among the calculation results of all simulation calculation processes.

[0155] Optionally, the machine learning model of the target water injection pump includes a first model, a second model, a third model, and a fourth model, wherein the target water injection pump is any candidate water injection pump, and the first model is a linear model;

[0156] The first model of the target water injection pump is pre-trained using the historical single pump frequency of the target water injection pump as the training input sample and the single pump flow rate of the target water injection pump when it is turned on at the historical single pump frequency as the training target.

[0157] The second model of the target water injection pump is pre-trained using historical water tank level and historical single pump flow rate as training input samples and the pump inlet pressure of the target water injection pump when it is running under the historical water tank level and historical single pump flow rate as the training target.

[0158] The third model of the target water injection pump is pre-trained using historical external pressure and historical single pump flow rate as training input samples, and the pump post-pump pressure of the target water injection pump when it is running under the historical water storage tank level and historical single pump flow rate as the training target.

[0159] The fourth model of the target water injection pump is pre-trained using the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as training input samples, and the energy consumption of the target water injection pump when operating under the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as the training target.

[0160] Optionally, the simulation calculation unit is used to obtain a candidate frequency set, including: the simulation calculation unit is specifically used for:

[0161] Obtain the single pump frequency of the first type of candidate water injection pump in the candidate pump start combination, where the number of the first type of candidate water injection pump is n-1, and n is equal to the number of candidate water injection pumps in the candidate pump start combination;

[0162] Based on the single pump frequency and the first model of each of the first type of candidate water injection pumps, the single pump flow rate of each of the first type of candidate water injection pumps is obtained.

[0163] The difference between the outflow rate and the single pump flow rate of all the first type of candidate water injection pumps is taken as the single pump flow rate of the second type of candidate water injection pump. The second type of candidate water injection pump is the candidate water injection pump other than the first type of candidate water injection pump in the candidate pump start combination.

[0164] Based on the single-pump flow rate of the second type of candidate water injection pumps and the second model, the single-pump frequency of the second type of candidate water injection pumps is obtained.

[0165] Obtain the candidate frequency set, which includes the single pump frequency of each of the first type of candidate water injection pumps and the single pump frequency of each of the second type of candidate water injection pumps.

[0166] Optionally, the simulation calculation unit is used to execute multiple simulation calculation processes, including: a simulation calculation unit specifically used to iteratively execute multiple simulation calculation processes;

[0167] The optimal energy consumption conditions also include: the difference between the calculation results of a series of simulation calculations is less than a preset difference threshold, and / or the number of iterations is greater than a preset number threshold.

[0168] Optionally, the device further includes:

[0169] An optimization unit is used to update the single pump frequency of each first-type candidate water injection pump using a preset optimization algorithm if the simulation calculation process does not meet the optimal energy consumption condition, with the minimum calculation result as the optimization condition, so as to obtain the update frequency of each first-type candidate water injection pump.

[0170] The simulation calculation unit is used to obtain the single pump frequency of the first type of candidate water injection pump in the candidate pump start-up combination, including: The simulation calculation unit is specifically used for:

[0171] If the simulation calculation process is the first simulation calculation process, the preset initial frequency of each of the first type of candidate water injection pumps is taken as the single pump frequency.

[0172] If the simulation calculation process is the kth iteration, the update frequency of each first-type candidate water injection pump obtained from the (k-1)th simulation calculation process is taken as the single pump frequency, where k is an integer greater than 1.

[0173] The simulation calculation unit is used to obtain the total energy consumption of the candidate pump start-up combination based on the demand parameters and the machine learning model of each candidate water injection pump. Specifically, the simulation calculation unit is used for:

[0174] The water level in the storage tank and the single pump flow rate of each candidate water injection pump are correspondingly input into the second model of each candidate water injection pump, and the output of the second model of each candidate water injection pump is used as the pump inlet pressure of each candidate water injection pump.

[0175] The external pressure and the single pump flow rate of each candidate water injection pump are input into the third model of each candidate water injection pump, and the output of the third model of each candidate water injection pump is used as the pump outlet pressure of each candidate water injection pump.

[0176] The single pump flow rate, single pump frequency, pump inlet pressure and pump outlet pressure of each candidate water injection pump are input into the fourth model of each candidate water injection pump, and the output of the fourth model of each candidate water injection pump is used as the energy consumption of each candidate water injection pump.

[0177] The sum of the energy consumption of each of the candidate water injection pumps is taken as the total energy consumption.

[0178] Optionally, the parameter acquisition unit is used to acquire candidate pump start combinations, which includes: acquiring an optimization set, the optimization set including multiple pump start combinations, each pump start combination including at least one water injection pump; and acquiring pump start combinations from the optimization set in a traversal manner as candidate pump start combinations.

[0179] The device also includes an optimal result determination unit, which is used to determine the optimal pumping parameters as the pumping parameters corresponding to the target candidate pumping combination if the calculation result of the simulation calculation process that satisfies the optimal energy consumption condition is the smallest.

[0180] The device for acquiring pump start-up parameters includes a processor and a memory. The parameter acquisition unit, model building unit, simulation calculation unit, and result acquisition unit are all stored as program units in the memory. The processor executes the program units stored in the memory to achieve the corresponding functions.

[0181] The processor contains a core, which retrieves the corresponding program units from memory. One or more cores can be configured; adjusting core parameters can improve the accuracy of pump-on parameters and reduce overall power consumption.

[0182] This invention provides a storage medium storing a program that, when executed by a processor, implements the method for obtaining pump start-up parameters.

[0183] This invention provides a processor for running a program, wherein the program executes the method for obtaining the pump start-up parameters during runtime.

[0184] This invention provides an electronic device, such as... Figure 6 As shown, the electronic device 60 includes at least one processor 601, at least one memory 602 connected to the processor, and a bus 603; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the above-described method for obtaining pump start-up parameters. The electronic device in this article can be a server, PC, PAD, mobile phone, etc.

[0185] This application also provides a computer program product, which, when executed on an electronic device, is suitable for executing a program that initializes the following method steps:

[0186] A method for obtaining pump start-up parameters, comprising:

[0187] Obtain candidate pump start combinations and preset demand parameters. The candidate pump start combinations include at least one candidate water injection pump. The demand parameters include the water tank level, the external pressure, and the external flow rate. The candidate water injection pump is the water injection pump in the water injection system.

[0188] A machine learning model for each candidate water injection pump is pre-trained based on historical data of the water injection system, wherein the historical data includes historical demand parameters and historical single pump frequency.

[0189] The simulation calculation process is executed multiple times. Each simulation calculation process includes: obtaining a candidate frequency set, which includes the single pump frequency of each candidate water injection pump; when the switch state of each candidate water injection pump is on, obtaining the total energy consumption of the candidate pump-on combination based on the demand parameters, the single pump frequency of each candidate water injection pump and the machine learning model, as the calculation result of the simulation calculation process.

[0190] If the simulation calculation process meets the preset optimal energy consumption conditions, the candidate pump start-up combination and the candidate frequency set corresponding to the simulation calculation process are used as pump start-up parameters. The optimal energy consumption conditions include the minimum calculation result among all simulation calculation process calculation results.

[0191] Optionally, the machine learning model of the target water injection pump includes a first model, a second model, a third model, and a fourth model, wherein the target water injection pump is any candidate water injection pump, and the first model is a linear model;

[0192] The first model of the target water injection pump is pre-trained using the historical single pump frequency of the target water injection pump as the training input sample and the single pump flow rate of the target water injection pump when it is turned on at the historical single pump frequency as the training target.

[0193] The second model of the target water injection pump is pre-trained using historical water tank level and historical single pump flow rate as training input samples and the pump inlet pressure of the target water injection pump when it is running under the historical water tank level and historical single pump flow rate as the training target.

[0194] The third model of the target water injection pump is pre-trained using historical external pressure and historical single pump flow rate as training input samples, and the pump post-pump pressure of the target water injection pump when it is running under the historical water storage tank level and historical single pump flow rate as the training target.

[0195] The fourth model of the target water injection pump is pre-trained using the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as training input samples, and the energy consumption of the target water injection pump when operating under the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as the training target.

[0196] Optionally, a candidate frequency set is obtained, including:

[0197] Obtain the single pump frequency of the first type of candidate water injection pump in the candidate pump start combination, where the number of the first type of candidate water injection pump is n-1, and n is equal to the number of candidate water injection pumps in the candidate pump start combination;

[0198] Based on the single pump frequency and the first model of each of the first type of candidate water injection pumps, the single pump flow rate of each of the first type of candidate water injection pumps is obtained.

[0199] The difference between the outflow rate and the single pump flow rate of all the first type of candidate water injection pumps is taken as the single pump flow rate of the second type of candidate water injection pump. The second type of candidate water injection pump is the candidate water injection pump other than the first type of candidate water injection pump in the candidate pump start combination.

[0200] Based on the single-pump flow rate of the second type of candidate water injection pumps and the second model, the single-pump frequency of the second type of candidate water injection pumps is obtained.

[0201] Obtain the candidate frequency set, which includes the single pump frequency of each of the first type of candidate water injection pumps and the single pump frequency of each of the second type of candidate water injection pumps.

[0202] Optionally, the simulation calculation process can be executed multiple times, including: iteratively executing the simulation calculation process multiple times;

[0203] The optimal energy consumption conditions also include: the difference between the calculation results of a series of simulation calculations is less than a preset difference threshold, and / or the number of iterations is greater than a preset number threshold.

[0204] Optionally, the method further includes:

[0205] If the simulation calculation process does not meet the optimal energy consumption condition, the minimum calculation result is used as the optimization condition, and the single pump frequency of each first type of candidate water injection pump is updated using a preset optimization algorithm to obtain the update frequency of each first type of candidate water injection pump.

[0206] The step of obtaining the single pump frequency of the first type of candidate water injection pump in the candidate pump start-up combination includes:

[0207] If the simulation calculation process is the first simulation calculation process, the preset initial frequency of each of the first type of candidate water injection pumps is taken as the single pump frequency.

[0208] If the simulation calculation process is the kth iteration, the update frequency of each first-type candidate water injection pump obtained from the (k-1)th simulation calculation process is taken as the single pump frequency, where k is an integer greater than 1.

[0209] Optionally, based on the demand parameters and the machine learning model of each candidate water injection pump, the total energy consumption of the candidate pump-starting combination is obtained, including:

[0210] The water level in the storage tank and the single pump flow rate of each candidate water injection pump are correspondingly input into the second model of each candidate water injection pump, and the output of the second model of each candidate water injection pump is used as the pump inlet pressure of each candidate water injection pump.

[0211] The external pressure and the single pump flow rate of each candidate water injection pump are input into the third model of each candidate water injection pump, and the output of the third model of each candidate water injection pump is used as the pump outlet pressure of each candidate water injection pump.

[0212] The single pump flow rate, single pump frequency, pump inlet pressure and pump outlet pressure of each candidate water injection pump are input into the fourth model of each candidate water injection pump, and the output of the fourth model of each candidate water injection pump is used as the energy consumption of each candidate water injection pump.

[0213] The sum of the energy consumption of each of the candidate water injection pumps is taken as the total energy consumption.

[0214] Optionally, obtaining candidate pump start combinations includes: obtaining an optimization set, the optimization set including multiple pump start combinations, each pump start combination including at least one water injection pump;

[0215] Pump start combinations are obtained from the optimization set in a traversal manner and used as candidate pump start combinations;

[0216] The method further includes:

[0217] If the simulation calculation result of the target candidate pump start combination that satisfies the optimal energy consumption condition is the smallest, the pump start parameters corresponding to the target candidate pump start combination are taken as the optimal pump start parameters.

[0218] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, electronic devices (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable device, generate instructions for implementing the flowchart... Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0219] In a typical configuration, an electronic device includes one or more processors (CPUs), memory, and a bus. The electronic device may also include input / output interfaces, network interfaces, etc.

[0220] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM, and memory includes at least one memory chip. Memory is an example of computer-readable media.

[0221] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0222] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0223] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0224] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for obtaining pump start-up parameters, characterized in that, include: Obtain candidate pump start combinations and preset demand parameters. The candidate pump start combinations include at least one candidate water injection pump. The demand parameters include the water tank level, the external pressure, and the external flow rate. The candidate water injection pump is the water injection pump in the water injection system. A machine learning model for each candidate water injection pump is pre-trained based on historical data of the water injection system, wherein the historical data includes historical demand parameters and historical single pump frequency. The simulation calculation process is executed multiple times. Each simulation calculation process includes: obtaining a candidate frequency set, which includes the single pump frequency of each candidate water injection pump; when the switch state of each candidate water injection pump is on, obtaining the total energy consumption of the candidate pump-on combination based on the demand parameters, the single pump frequency of each candidate water injection pump and the machine learning model, as the calculation result of the simulation calculation process. If the simulation calculation process meets the preset optimal energy consumption conditions, the candidate pump start-up combination and the candidate frequency set corresponding to the simulation calculation process are used as pump start-up parameters. The optimal energy consumption conditions include the minimum calculation result among all simulation calculation process calculation results.

2. The method according to claim 1, characterized in that, The machine learning model for the target water injection pump includes a first model, a second model, a third model, and a fourth model, wherein the target water injection pump is any candidate water injection pump, and the first model is a linear model; The first model of the target water injection pump is pre-trained using the historical single pump frequency of the target water injection pump as the training input sample and the single pump flow rate of the target water injection pump when it is turned on at the historical single pump frequency as the training target. The second model of the target water injection pump is pre-trained using historical water tank level and historical single pump flow rate as training input samples and the pump inlet pressure of the target water injection pump when it is running under the historical water tank level and historical single pump flow rate as the training target. The third model of the target water injection pump is pre-trained using historical external pressure and historical single pump flow rate as training input samples, and the pump post-pump pressure of the target water injection pump when it is running under the historical water storage tank level and historical single pump flow rate as the training target. The fourth model of the target water injection pump is pre-trained using the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as training input samples, and the energy consumption of the target water injection pump when operating under the historical single pump flow rate, historical pump inlet pressure, historical pump outlet pressure, and historical single pump frequency as the training target.

3. The method according to claim 2, characterized in that, The acquisition of the candidate frequency set includes: Obtain the single pump frequency of the first type of candidate water injection pump in the candidate pump start combination, where the number of the first type of candidate water injection pump is n-1, and n is equal to the number of candidate water injection pumps in the candidate pump start combination; Based on the single pump frequency and the first model of each of the first type of candidate water injection pumps, the single pump flow rate of each of the first type of candidate water injection pumps is obtained. The difference between the outflow rate and the single pump flow rate of all the first type of candidate water injection pumps is taken as the single pump flow rate of the second type of candidate water injection pump. The second type of candidate water injection pump is the candidate water injection pump other than the first type of candidate water injection pump in the candidate pump start combination. Based on the single-pump flow rate of the second type of candidate water injection pumps and the second model, the single-pump frequency of the second type of candidate water injection pumps is obtained. Obtain the candidate frequency set, which includes the single pump frequency of each of the first type of candidate water injection pumps and the single pump frequency of each of the second type of candidate water injection pumps.

4. The method according to claim 3, characterized in that, The process of performing multiple simulation calculations includes: iteratively performing multiple simulation calculations. The optimal energy consumption conditions also include: the difference between the calculation results of a series of simulation calculations is less than a preset difference threshold, and / or the number of iterations is greater than a preset number threshold.

5. The method according to claim 4, characterized in that, The method further includes: If the simulation calculation process does not meet the optimal energy consumption condition, the minimum calculation result is used as the optimization condition, and the single pump frequency of each first type of candidate water injection pump is updated using a preset optimization algorithm to obtain the update frequency of each first type of candidate water injection pump. The step of obtaining the single pump frequency of the first type of candidate water injection pump in the candidate pump start-up combination includes: If the simulation calculation process is the first simulation calculation process, the preset initial frequency of each of the first type of candidate water injection pumps is taken as the single pump frequency. If the simulation calculation process is the kth iteration, the update frequency of each first-type candidate water injection pump obtained from the (k-1)th simulation calculation process is taken as the single pump frequency, where k is an integer greater than 1.

6. The method according to claim 3 or 5, characterized in that, The step of obtaining the total energy consumption of the candidate pump-starting combinations based on the demand parameters and the machine learning model of each candidate water injection pump includes: The water level in the storage tank and the single pump flow rate of each candidate water injection pump are correspondingly input into the second model of each candidate water injection pump, and the output of the second model of each candidate water injection pump is used as the pump inlet pressure of each candidate water injection pump. The external pressure and the single pump flow rate of each candidate water injection pump are input into the third model of each candidate water injection pump, and the output of the third model of each candidate water injection pump is used as the pump outlet pressure of each candidate water injection pump. The single pump flow rate, single pump frequency, pump inlet pressure and pump outlet pressure of each candidate water injection pump are input into the fourth model of each candidate water injection pump, and the output of the fourth model of each candidate water injection pump is used as the energy consumption of each candidate water injection pump. The sum of the energy consumption of each of the candidate water injection pumps is taken as the total energy consumption.

7. The method according to claim 1, characterized in that, The process of obtaining candidate pump start combinations includes: obtaining an optimization set, wherein the optimization set includes multiple pump start combinations, and each pump start combination includes at least one water injection pump. Pump start combinations are obtained from the optimization set in a traversal manner and used as candidate pump start combinations; The method further includes: If the simulation calculation result of the target candidate pump start combination that satisfies the optimal energy consumption condition is the smallest, the pump start parameters corresponding to the target candidate pump start combination are taken as the optimal pump start parameters.

8. A device for acquiring pump start-up parameters, characterized in that, include: The parameter acquisition unit is used to acquire candidate pump start combinations and preset demand parameters. The candidate pump start combinations include at least one candidate water injection pump. The demand parameters include the water tank level, the external pressure, and the external flow rate. The candidate water injection pump is a water injection pump in the water injection system. The model building unit is used to pre-train a machine learning model for each candidate water injection pump based on historical data of the water injection system. The historical data includes historical demand parameters and historical single pump frequency. The simulation calculation unit is used to execute multiple simulation calculation processes. Each simulation calculation process includes: obtaining a candidate frequency set, which includes the single pump frequency of each candidate water injection pump; when the switching state of each candidate water injection pump is on, obtaining the total energy consumption of the candidate pump-on combination based on the demand parameters, the single pump frequency of each candidate water injection pump, and the machine learning model, as the calculation result of the simulation calculation process. The result acquisition unit is used to take the candidate pump start combination and the candidate frequency set corresponding to the simulation calculation process as pump start parameters if the simulation calculation process meets the preset optimal energy consumption conditions. The optimal energy consumption conditions include the minimum calculation result among the calculation results of all simulation calculation processes.

9. A storage medium, characterized in that, The storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the storage medium to perform the method for obtaining pump start parameters as described in any one of claims 1-7.

10. An electronic device, comprising at least one processor, and at least one memory and a bus connected to the processor; wherein, The processor and the memory communicate with each other via the bus; The processor is used to call program instructions in the memory to execute the method for obtaining pump start-up parameters as described in any one of claims 1-7.

Images