Water supply control method based on multi-redundant neural network

Abstract

The invention discloses a water supply control method based on a multi-redundant neural network, and relates to the technical field of water supply control. The technical key points are as follows: the system collects feedback data of each key node of the water supply system in real time, and carries out deep analysis on an operation attenuation coefficient and a state estimation coefficient by using a multi-redundancy neural network algorithm, so that subtle changes in the operation process of execution equipment can be accurately captured, potential fault hidden dangers can be timely perceived, and once abnormality is detected, the system can be used for monitoring the operation of the execution equipment. The system rapidly triggers an emergency plan, automatically switches to a standby model to ensure the stability of water supply, dynamically adjusts a water supply strategy according to the fault type, immediately sends alarm information to a manager, explains the fault position, the fault type and the possible influence in detail, and comprehensively avoids the serious interference of the fault on water supply.

Description

Technical Field

[0001] The present invention relates to the technical field of water supply control, and in particular to a water supply control method based on a multi-redundant neural network. Background Art

[0002] With the continuous expansion of urban scale and the steady improvement of residents' quality of life, the stability and reliability of water supply systems face extremely severe challenges. Traditional water supply control modes rely on simple mathematical models or past experience rules, which are powerless in the face of complex and changeable water supply demands and the dynamic operation of the pipeline network. For example, during peak water use periods, the water supply pressure often fails to meet demand; and during low water use periods, it is easy to cause unnecessary waste of energy. In addition, once the internal equipment of the pipeline network fails, it is very likely to cause serious problems such as water supply interruption or water quality deterioration, which will have a great impact on the daily lives of residents.

[0003] At present, a Chinese patent with the existing patent application number "CN202410264917.6" discloses a water supply control method based on a multi-redundant neural network. It includes a master system, multiple slave systems and a frequency converter. The frequency converter is set corresponding to the number of slave systems. The master system and the slave system are dually connected through the RJ45 Ethernet interface and the I / O port. The master system and each frequency converter are connected through the I / O port; the slave systems are dually connected through the RJ45 Ethernet interface and the I / O port; the slave system and the frequency converter are dually connected through the RS485 interface and the I / O port; the master system, the slave system, and the frequency converter receive the water outlet pressure signal through the I / O port respectively; although the invention forms a three-level control operation system by setting the master system, the slave system, and the frequency converter. To achieve the effectiveness and safety of the water supply system operation, it seriously ignores the management of the water supply equipment, so that the water supply equipment fails, shorts, or is even completely damaged, and it cannot be understood and diagnosed in time.

[0004] It is impossible to effectively supervise water supply equipment; traditional water supply control methods mostly rely on fixed parameter settings and empirical control strategies, which are inadequate in the face of complex and ever-changing demands and operating conditions in the water supply system. For example, water supply equipment operates at high load for a long time, which is prone to sudden mechanical failures, such as wear of the pump impeller and motor short circuit; during the peak period of urban water use, the demand for residential water and commercial water is superimposed, and the water consumption is exploding; the pipeline has been eroded for many years, and is affected by geological subsidence and external force compression, which may cause leakage, rupture and other conditions; in all the above situations, traditional control methods are subject to their own limitations, and it is difficult to capture system anomalies in time, and the control action is slow and the accuracy is poor. It is impossible to quickly and appropriately adjust key parameters such as water supply flow and pressure, resulting in unstable water pressure and water volume problems in water supply, which seriously interferes with the normal water use order of users, brings many inconveniences to residents' lives, and also buries safety hazards for industrial production. Summary of the invention

[0005] 1. Technical issues to be resolved In view of the deficiencies of the prior art, the present invention provides a water supply control method based on a multi-redundant neural network, which solves the problems raised in the background technology.

[0006] (II) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions: A water supply control method based on a multi-redundant neural network, the method comprising the following steps: Sensors are deployed at key nodes of the water supply system, such as water sources, pumping stations, reservoirs, and user terminals, to collect real-time data such as water pressure, flow, water quality, and equipment operating status. The collected data is denoised using a filtering algorithm, and the data is normalized to the [0,1] interval using a normalization formula to complete data preprocessing.

[0007] Multi-redundant neural network construction link: Construct multiple parallel neural network models, each of which has an input layer, a hidden layer, and an output layer. The input layer sets the corresponding number of neurons according to the type of data collected; the hidden layer extracts features and recognizes patterns of data through specific neuron combinations and weight configurations; the output layer outputs water supply control related parameters such as pump speed, valve opening, etc., and a redundant and complementary relationship is formed between the neural network models.

[0008] Model training and optimization: Use historical data of the water supply system and actual water supply control effect data to train multiple redundant neural networks. Use optimization algorithms such as back propagation algorithms to continuously adjust the weights and thresholds of the neural network model, and regularly update training data to adapt to the dynamic changes in the operation of the water supply system.

[0009] Real-time control decision-making link: The real-time collected and pre-processed data is input into the trained multi-redundant neural network. Each neural network model synchronously outputs the water supply control parameters, and the output results of multiple neural network models are comprehensively analyzed and judged through the decision fusion algorithm (including but not limited to the weighted average algorithm, voting algorithm, etc.), so as to determine the optimal water supply control plan and send the control instructions to the corresponding execution device.

[0010] Fault diagnosis and processing: Real-time monitoring of the operating status of multiple redundant neural networks, by comparing the output results of each neural network model and comparing them with the preset normal range, to determine whether there is a model failure or water supply system failure. Once a fault is detected, the backup model is immediately started, or corresponding processing measures such as adjusting the control strategy and issuing an alarm signal are taken.

[0011] Furthermore, the key nodes for data collection include water sources, pumping stations, reservoirs, water pipe nodes and user ends. The data types collected include water pressure, flow, water quality and equipment operating status. After denoising, the data is normalized to the [0, 1] interval using the normalization formula to complete data preprocessing.

[0012] Furthermore, before calculating the stability assessment index, it is necessary to analyze the output ratio of the boosting equipment in the designated area. The analysis process is as follows: Set the area division range and obtain the water pressure of each key node of the water supply system in the area ,flow And the operating data of the booster equipment, where the flow rate is in cubic meters per second, and the operating data of the booster equipment includes the output power , Daily usage time And daily electricity consumption , output power The unit is kilowatt-hour, daily usage time The unit is hour, daily power consumption The unit is kilowatt-hour; When the output power of the booster device and daily usage time Compare with the corresponding evaluation stability interval respectively. When the output power Or daily usage time If it is not within the corresponding evaluation stability interval, the data of the supercharging equipment is discarded and a maintenance instruction is issued to the maintenance department; otherwise, no response is made, wherein the evaluation stability interval includes an upper threshold and a lower threshold; By analyzing the formula Get the output proportionality coefficient of the booster device , where N represents the number of all booster equipment in the designated area, represents the average daily power consumption of the i-th booster equipment, Indicates the daily power consumption of the current booster equipment. Indicates the output power of the current booster device. , They represent the upper and lower output power thresholds of the current boost device. represents the output power of the ith booster device, Indicates the daily usage time of the current booster equipment. , They represent the upper and lower daily usage time thresholds of the current booster equipment, respectively. represents the daily usage time of the i-th booster equipment, Respectively represent the weights of the preset output power, daily power consumption and daily usage time, among which, when the output proportional coefficient ≥1, the output proportional coefficient calculated this time is output Otherwise, the output proportional coefficient calculated this time will be discarded. .

[0013] Furthermore, the formula for calculating the stability assessment index is as follows:

[0014] In the formula, The stability assessment index representing the stability in the zoned area, represents the correction coefficient of the preset stability assessment index, e represents a natural constant, A represents a constant, and 5≥A>3, represents the average output proportional coefficient of the supercharging equipment in the designated area, M represents the number of key nodes in the designated area, represents the water pressure value of the jth key node, , Respectively represent the maximum flow and minimum flow in each key node, represent the preset average output proportional coefficient, water pressure and flow weights respectively, and , It represents the average output proportional coefficient of the repaired booster equipment in the zoned area. Indicates the maintenance time interval of the booster equipment.

[0015] Furthermore, the obtained output proportional coefficient is compared with the average output proportional coefficient of the supercharging equipment in the zoned area. >85%, no response. When ≤85%, a pressure regulation command is issued, and the output proportional coefficients of the two boosting devices before and after the boosting device in the water supply system are obtained, which are recorded as and ,like ≤85% or ≤85%, a maintenance order will be issued to repair the voltage stabilizing equipment; if >85% or >85%, the voltage regulation instruction is executed to adjust the and Output power of the corresponding booster equipment; Among them, when the booster equipment is adjusted, As the adjustment ratio of output power.

[0016] Furthermore, the decision fusion algorithm includes but is not limited to a weighted average algorithm and a voting algorithm, which determines the final water supply control plan by fusing the output results of multiple neural network models. After confirming the optimal plan, it is achieved by comparing the output results of multiple neural network models and comparing the output results with a preset normal range to determine whether there is a fault in the model or the water supply system.

[0017] Furthermore, the analysis process of running the attenuation coefficient is as follows: Set the time interval for data retrieval, recorded as , get the current and The output proportional coefficient of the same booster equipment and the stability assessment index of the area where it is located days ago; By analyzing the formula Get running time Operational attenuation factor , represents the correction factor of the operating attenuation coefficient, where: , Respectively represent the current output proportional coefficient and Output scale factor days ago, , Respectively represent the current stability assessment index of the area where the same booster equipment is located and Stability assessment index days ago; If the operating attenuation factor When the preset balance threshold is exceeded, a maintenance check instruction is issued to the maintenance department, and the operation attenuation coefficient is recalculated based on the actual test data. The correction coefficient is adjusted based on the difference between the two calculations. otherwise, no response.

[0018] Furthermore, the two calculated running attenuation coefficients are compared. When the difference between the two is within the preset error tolerance threshold range, a qualified instruction is issued and the correction coefficient is not adjusted. When the difference between the two is greater than the upper limit threshold in the error tolerance threshold range, an upward adjustment instruction is issued; otherwise, a downward adjustment instruction is issued; When receiving an upward command, As the correction factor of the corrected running attenuation coefficient; when receiving a downward adjustment instruction, As the correction factor for the corrected operating attenuation factor, ∈(0.12, 0.15).

[0019] Furthermore, the formula for calculating the state estimation coefficient is as follows:

[0020] In the formula, represents the state estimation coefficient of the kth booster equipment in the divided area, represents the operating attenuation coefficient of the i-th booster equipment, , They represent the maximum and minimum operating attenuation coefficients of the booster equipment in the divided area, respectively. They represent the weights of the preset operation attenuation coefficient and output proportional coefficient respectively. ; when ≤1, then Recorded as 1; when > When , the state estimation coefficient obtained in this calculation is discarded, and the boosting equipment corresponding to the state estimation coefficient is marked.

[0021] Furthermore, when the state prediction coefficient exceeds the preset threshold, an alarm is issued, and the water supply in each period of the previous week is obtained. The period with the minimum water supply is selected as the maintenance time, and the boosting equipment in the area is repaired; otherwise, the current state continues to be monitored, relevant data is recorded, and the operating parameters of the boosting equipment are adjusted according to the real-time water supply.

[0022] (III) Beneficial effects The present invention provides a water supply control method based on a multi-redundant neural network, which has the following beneficial effects: Timely detection and handling of faults: The system obtains feedback data from key nodes of the water supply system in real time, and uses a multi-redundant neural network algorithm to conduct in-depth analysis of the operation attenuation coefficient and the state estimation coefficient. In this way, it is possible to keenly capture subtle changes in the operation of the execution equipment and detect potential faults in a timely manner. Once an abnormality is detected, the system will quickly activate the emergency plan and automatically switch to the backup model to maintain stable water supply. At the same time, it will dynamically adjust the water supply strategy according to the type of fault, and immediately send an alarm message to the management personnel, detailing the location, type and possible impact of the fault, to avoid serious impact of the fault on the water supply in all aspects.

[0023] Optimize energy utilization: In-depth analysis of the output ratio of boosting equipment under different working conditions, with the help of big data analysis and intelligent algorithms, intelligent control of boosting equipment operating parameters on the premise of accurately meeting water supply needs. By optimizing equipment start and stop time, adjusting operating frequency and other means, the operating mode of boosting equipment is continuously optimized, energy utilization efficiency is effectively improved, energy consumption is effectively reduced, and the water supply system is helped to achieve green and energy-saving operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is the overall structural process of the present invention. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0026] In a certain city's water supply system, a variety of sensors are deployed in all directions to achieve accurate monitoring. At the water source, the water pressure sensor constantly monitors the changes in water source pressure, and the water quality sensor performs real-time detection of water source quality to ensure the purity and safety of the water source. At the pump station, the flow sensor accurately measures the water flow, and the equipment operation status sensor closely monitors the operation of the pump station equipment and promptly detects potential failure risks. The reservoir is equipped with a liquid level sensor to monitor the water level, and the water pressure sensor controls the water pressure status inside the reservoir. At the water pipe node, the flow sensor and the water pressure sensor work together to grasp the flow and pressure distribution of the water flow in the pipe network. At the user end, the water meter records the user's water consumption in detail. These sensors are like the "tentacles" of the water supply system, collecting various types of data in real time. The collected data will first pass through the filtering algorithm to effectively remove the noise interference caused by environmental interference and other factors, making the data purer and more accurate. Next, the normalization formula is used to normalize different types of data such as water pressure, flow, water quality, and equipment operating status to the interval [0, 1], thereby fully preparing for the input of the subsequent neural network model and ensuring that the data format is unified and meets the model requirements.

[0027] 1. Example of building and training a multi-redundant neural network model Five parallel neural network models are constructed to enhance the stability and accuracy of the system. The input layer sets the number of neurons according to the data type. For example, in view of the importance and complexity of water pressure data, five neurons are set to process water pressure data so that it can fully capture the changing characteristics of water pressure; three neurons are set for flow data to accurately process flow-related information. The hidden layer uses 50 neurons to deeply mine the input data through activation functions, such as the ReLU function, to effectively extract data features and identify complex patterns, laying the foundation for the accurate output of the subsequent output layer. The output layer outputs water supply control parameters, such as the output power of the booster equipment and the opening of the valve. These parameters directly determine the operating status of the water supply system. Using the historical data of the city's water supply system in the past year and the actual control effect data, the back propagation algorithm is used to train the five neural network models. In order to better adapt to various changes that may occur in the water supply system, such as seasonal changes in user water consumption, large water consumption in summer and relatively small water consumption in winter, and the gradual decline of equipment performance over time, the training data is updated every two weeks so that the model can keep pace with the times and always maintain accurate control of the water supply system.

[0028] Key nodes and data types: Sensors are deployed at the water source, pump station, reservoir, water pipe node and user end of the water supply system to collect data such as water pressure, flow, water quality and equipment operation status. The collected data is denoised using a filtering algorithm, and then the data is normalized to the [0,1] interval using a normalization formula to complete data preprocessing.

[0029] 2. Implementation example of stability assessment index calculation and water supply system adjustment Set a certain area as the research object, and use the sensor network to fully obtain the water pressure, flow rate and operation data of the booster equipment at each key node of the water supply system in the area. For example, the output power of a booster device is 100 kWh, and the daily usage time is 8 hours. These data are carefully compared with the corresponding evaluation stability interval (output power 80-120 kWh, daily usage time 6-10 hours). After comparison, it is found that all the data of the booster device are within the stable range, so the system does not respond temporarily. Subsequently, the output proportional coefficient of the booster device is calculated by a specific analysis formula. Assuming that the calculated output proportional coefficient is 1.2, the system will output the coefficient. Then, according to the stability evaluation index calculation formula, the stability evaluation index of the area is calculated by comprehensively considering multiple factors such as water pressure, flow rate, and equipment operation time. If the calculation result shows that the stability evaluation index is low, it indicates that the water supply stability in the area needs to be improved. At this time, the system will adjust the water supply system according to the pre-established relevant rules, and appropriately adjust the output power of the booster equipment to improve the stability of water supply and ensure the normal water use of users in the area.

[0030] 2.1. Set the area division range and obtain the water pressure of each key node of the water supply system in the area. ,flow And the operating data of the booster equipment, where the flow rate is in cubic meters per second, and the operating data of the booster equipment includes the output power , Daily usage time And daily electricity consumption , output power The unit is kilowatt-hour, daily usage time The unit is hour, daily power consumption The unit is kilowatt-hour.

[0031] Furthermore, when the output power of the booster device and daily usage time Compare with the corresponding evaluation stability interval respectively. When the output power Or daily usage time If it is not within the corresponding evaluation stability interval, the data of the supercharging equipment is discarded and a maintenance instruction is issued to the maintenance department; otherwise, no response is made, wherein the evaluation stability interval includes an upper threshold and a lower threshold.

[0032] 2.2. Output proportional coefficient

[0033] By using the water supply system data of the area in question and combining it with historical operation records, we analyze its changing trends and further optimize the operation strategy of the booster equipment to ensure the long-term stable operation of the water supply system. At the same time, we monitor data changes in real time and adjust parameters in a timely manner to prevent abnormal situations. In this way, the system can not only detect potential problems in a timely manner, but also effectively prevent sudden failures of the water supply system and ensure the safety of users' water use. In addition, regular in-depth analysis of data can help improve overall water supply efficiency, reduce energy consumption, and achieve optimal resource allocation.

[0034] Output scale factor The formula is as follows:

[0035] In the formula, N—represents the number of all boosting equipment in the designated area; — represents the average daily power consumption of the i-th booster equipment; —Indicates the daily power consumption of the current booster equipment; —Indicates the output power of the current booster equipment; , —represent the output power upper limit threshold and output power lower limit threshold of the current boost device respectively; — represents the output power of the i-th booster device; —Indicates the daily usage time of the current booster equipment; , —represent the upper and lower thresholds of the daily usage time of the current boosting equipment respectively; — represents the daily usage time of the i-th booster equipment; —represent the preset output power, daily power consumption and daily usage time weights respectively, among which, when the output proportional coefficient ≥1, the output proportional coefficient calculated this time is output Otherwise, the output proportional coefficient calculated this time will be discarded. .

[0036] Compare the obtained output proportional coefficient with the average output proportional coefficient of the supercharging equipment in the zoned area. >85%, no response. When ≤85%, a pressure regulation command is issued, and the output proportional coefficients of the two boosting devices before and after the boosting device in the water supply system are obtained, which are recorded as and ,like ≤85% or ≤85%, a maintenance instruction is issued to repair the voltage stabilizing equipment; if >85% or >85%, the voltage regulation instruction is executed to adjust the and Output power of the corresponding booster equipment; Among them, when the booster equipment is adjusted, As the adjustment ratio of output power.

[0037] Through this calculation mechanism, the system can accurately identify high-energy consumption equipment, adjust the operation strategy in time, ensure that the water supply system strikes a balance between high efficiency and energy saving, further optimize resource allocation, and improve overall operation efficiency. In this way, while ensuring the stability of water supply, it can significantly reduce operating costs, achieve the goal of sustainable development, and lay a solid foundation for the intelligent upgrade of urban water supply systems.

[0038] 2.3 Analysis of Stability Assessment Index The stability assessment index quantifies system stability by comprehensively considering equipment operation data, providing a scientific basis for decision-making. The index combines historical and real-time data to dynamically monitor equipment status, warn of potential risks, ensure the long-term stable operation of the water supply system, improve emergency response capabilities, and ensure the safety of urban water supply. The application of the stability assessment index not only improves the predictability of equipment maintenance, but also optimizes resource allocation and reduces the frequency of sudden failures. Through real-time data analysis, the system can accurately locate the source of the problem and respond quickly, effectively shortening the fault handling time, further enhancing the reliability and safety of the water supply system, and providing urban residents with more stable and efficient water supply services. The details are as follows:

[0039] In the formula, —Stability assessment index indicating stability within the zoned area; —Indicates the correction factor of the preset stability assessment index; e—represents a natural constant, A—represents a constant, and 5≥A>3; —Indicates the average output proportional coefficient of the booster equipment in the zoned area; M—represents the number of key nodes in the area. — represents the water pressure value of the jth key node, , —represents the maximum flow and minimum flow in each key node respectively; —represent the preset average output proportional coefficient, water pressure and flow weights respectively, and ; —Indicates the average output ratio coefficient of the repaired booster equipment in the zoned area; —Indicates the maintenance time interval of the boosting equipment.

[0040] By obtaining the stable evaluation index in the designated area, combined with the preset correction coefficient and real-time data, the system dynamically adjusts the equipment operation strategy to ensure the balance of water pressure and flow at each key node, further improving the overall stability of the water supply system. The reasonable setting of the maintenance time interval effectively reduces the equipment failure rate and ensures long-term efficient operation.

[0041] 3. Decision-making and Execution Example The adjusted data is input into the five trained multi-redundant neural network models, and each model outputs control parameters based on its own training results and data processing methods. For example, model 1 outputs the boosting equipment output power of 110 kWh, model 2 outputs 105 kWh, and so on. In order to obtain more accurate and reliable control instructions, a weighted average algorithm is used for decision fusion. When setting the weights, the reliability of each model and its performance in different scenarios are fully considered. After precise calculation, the final boosting equipment output power is 108 kWh. The system sends the instruction to the execution device to accurately control the operation of the boosting equipment and ensure that the water supply system operates stably according to the optimized parameters.

[0042] The decision fusion algorithm includes but is not limited to the weighted average algorithm and the voting algorithm. The final water supply control plan is determined by fusing the output results of multiple neural network models. After confirming the optimal plan, it is achieved by comparing the output results of multiple neural network models and comparing the output results with the preset normal range to determine whether there is a fault in the model or the water supply system.

[0043] 4. Feedback and troubleshooting implementation examples Real-time acquisition of water supply system feedback data. In order to grasp the system operation status in time, the data retrieval time interval is set to 1 day. Through this mechanism, the output proportional coefficient of the same booster equipment and the stability evaluation index of the area where it is located are obtained at present and 1 day ago, and then the operation attenuation coefficient is calculated through a special analysis formula. If the operation attenuation coefficient exceeds the preset balance threshold, it means that the equipment operation may be abnormal, and the system will issue a maintenance check instruction to the maintenance department. After the maintenance personnel arrive at the site, they recalculate the operation attenuation coefficient based on the data obtained from the actual test. Assuming that the operation attenuation coefficient is 0.8 for the first calculation and 0.9 for the recalculation, the difference between the two is 0.1. After judging that the difference is within the preset error allowable threshold range, the system issues a qualified instruction and does not adjust the correction coefficient. If the state estimation coefficient exceeds the preset threshold, the system immediately issues an alarm. In order to more accurately determine the cause of the fault, the system will obtain the water supply in each period of the previous week. After data analysis, it is found that the water supply is the smallest at 2-4 am. Based on this, the booster equipment in the area is repaired during this period to minimize the impact on user water use.

[0044] 4.1. Analysis of operation attenuation coefficient S001. Set the time interval for data retrieval, recorded as , get the current and The output proportional coefficient of the same booster equipment and the stability assessment index of the area where it is located days ago; S002, by analyzing the formula Get running time Operational attenuation factor , represents the correction factor of the operating attenuation coefficient, where: , Respectively represent the current output proportional coefficient and Output scale factor days ago, , Respectively represent the current stability assessment index of the area where the same booster equipment is located and Stability assessment index days ago; S003, if the running attenuation coefficient When the preset balance threshold is exceeded, a maintenance check instruction is issued to the maintenance department, and the operation attenuation coefficient is recalculated based on the actual test data. The correction coefficient is adjusted based on the difference between the two calculations. otherwise, no response.

[0045] S004. Compare the two calculated running attenuation coefficients. When the difference between the two is within the preset error tolerance threshold range, issue a qualified instruction and do not modify the correction coefficient. When the difference between the two is greater than the upper limit threshold in the error tolerance threshold range, an upward adjustment instruction is issued; otherwise, a downward adjustment instruction is issued; Among them, when receiving the upward instruction, As the correction factor of the corrected running attenuation coefficient; when receiving a downward adjustment instruction, As the correction factor for the corrected operating attenuation factor, ∈(0.12, 0.15).

[0046] 4.2 State Estimation Coefficient According to the trend of historical data changes and the current equipment operation status, the estimated coefficient is adjusted dynamically to ensure the calculation accuracy. By real-time monitoring of data changes, the correction strategy is optimized to improve the stability of equipment operation. Combining historical data with real-time monitoring, the estimated coefficient is adjusted dynamically, and the correction strategy is accurately optimized to effectively improve the stability of equipment operation and ensure the long-term and efficient operation of the system. Combined with real-time monitoring data, the estimated coefficient is dynamically fine-tuned to further optimize the correction strategy to ensure that the equipment operates in the best state, extend the service life of the equipment, and improve the overall system operation efficiency. The formula for calculating the state estimation coefficient is as follows:

[0047] In the formula, — represents the state estimation coefficient of the kth booster equipment in the divided area; — represents the operating attenuation coefficient of the i-th booster equipment; , —respectively represent the maximum operating attenuation coefficient and the minimum operating attenuation coefficient of the boosting equipment in the divided area; —represent the weights of the preset operation attenuation coefficient and output proportional coefficient, respectively, .

[0048] when ≤1, then Recorded as 1; when > When , the state estimation coefficient obtained in this calculation is discarded, and the boosting equipment corresponding to the state estimation coefficient is marked.

[0049] In contrast to the above, when the state prediction coefficient exceeds the preset threshold, an alarm is issued, and the water supply in each period of the previous week is obtained. The period with the minimum water supply is selected as the maintenance time, and the boosting equipment in the area is repaired; otherwise, the current state continues to be monitored, relevant data is recorded, and the operating parameters of the boosting equipment are adjusted according to the real-time water supply.

[0050] The weight coefficient is determined by the coefficient of variation method, which is a method of weighting each indicator according to the degree of variation between the current value of each evaluation indicator and the target value. If the numerical difference of an indicator is large, it can clearly distinguish the evaluated objects, indicating that the indicator has rich discrimination information, and thus the indicator should be given a larger weight. On the contrary, if the numerical difference of each evaluated object on a certain indicator is small, then the ability of this indicator to distinguish the evaluation objects is weak, and thus the indicator should be given a smaller weight. This method directly uses the information contained in each indicator and obtains the weight of the indicator through calculation, so it is objective.

[0051] In the application, the several formulas involved are all calculated by taking their numerical values ​​after removing the dimensions, and the formulas are established by collecting a large amount of data and performing software simulation to obtain a formula for the most recent actual situation. Some coefficients or weights in the formulas are set by technical personnel in this field according to actual conditions, so they will not be elaborated here.

[0052] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0053] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0054] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A water supply control method based on a multi-redundant neural network, characterized in that: The control method steps are as follows: S001. Deploy sensors at key nodes of the water supply system to collect real-time data and use filtering algorithms to remove noise; S002. Construct multiple parallel neural network models, each model has an input layer, a hidden layer and an output layer, wherein the input layer sets the number of neurons according to the data type, the hidden layer extracts features and recognizes patterns, and the output layer outputs water supply control parameters, and the models are redundant and complementary; S003. Use the historical and actual control effect data of the water supply system to train a multi-redundant neural network with a back propagation algorithm, regularly update the data to adapt to system changes, and use the control effect data to calculate the stability evaluation index of water supply in the area, and then adjust the water supply system based on the stability evaluation index; S004. Input the adjusted data into the trained multi-redundant neural network, each model outputs the control parameters, and the optimal solution is determined by the decision fusion algorithm, and the instructions are sent to the execution device, mainly the booster device; S005. Obtain feedback data from the water supply system in real time, analyze and execute the equipment operation attenuation coefficient, calibrate the operation parameters based on this to obtain the state estimation coefficient, compare it with the preset threshold to evaluate whether the equipment is faulty, and start the backup equipment or repair it if there is a fault.

2. The water supply control method based on multiple redundant neural networks according to claim 1 is characterized in that: The key nodes for data collection include water sources, pumping stations, reservoirs, water pipe nodes and user ends. The data types collected include water pressure, flow, water quality and equipment operating status. After denoising, the normalization formula is used to normalize the data to the [0, 1] interval to complete data preprocessing.

3. The water supply control method based on multiple redundant neural networks according to claim 1 is characterized in that: Before calculating the stability assessment index, it is necessary to analyze the output ratio of the booster equipment in the designated area. The analysis process is as follows: Set the area division range and obtain the water pressure of each key node of the water supply system in the area ,flow And the operating data of the booster equipment, where the flow rate is in cubic meters per second, and the operating data of the booster equipment includes the output power , Daily usage time And daily electricity consumption , output power The unit is kilowatt-hour, daily usage time The unit is hour, daily power consumption The unit is kilowatt-hour; When the output power of the booster device and daily usage time Compare with the corresponding evaluation stability interval respectively. When the output power Or daily usage time If it is not within the corresponding evaluation stability interval, the data of the supercharging equipment is discarded and a maintenance instruction is issued to the maintenance department; otherwise, no response is made, wherein the evaluation stability interval includes an upper threshold and a lower threshold; By analyzing the formula Get the output proportionality coefficient of the booster device , where N represents the number of all booster equipment in the designated area, represents the average daily power consumption of the i-th booster equipment, Indicates the daily power consumption of the current booster equipment. Indicates the output power of the current booster device. , They represent the upper and lower output power thresholds of the current boost device. represents the output power of the ith booster device, Indicates the daily usage time of the current booster equipment. , They represent the upper and lower daily usage time thresholds of the current booster equipment, respectively. represents the daily usage time of the i-th booster equipment, Respectively represent the weights of the preset output power, daily power consumption and daily usage time, among which, when the output proportional coefficient ≥1, the output proportional coefficient calculated this time is output Otherwise, the output proportional coefficient calculated this time will be discarded. .

4. The water supply control method based on multiple redundant neural networks according to claim 3 is characterized in that: The formula for calculating the stability assessment index is as follows: ； In the formula, The stability assessment index representing the stability in the zoned area, represents the correction coefficient of the preset stability assessment index, e represents a natural constant, A represents a constant, and 5≥A>3, represents the average output proportional coefficient of the supercharging equipment in the designated area, M represents the number of key nodes in the designated area, represents the water pressure value of the jth key node, , Respectively represent the maximum flow and minimum flow in each key node, represent the preset average output proportional coefficient, water pressure and flow weights respectively, and , It represents the average output proportional coefficient of the repaired booster equipment in the zoned area. Indicates the maintenance time interval of the booster equipment.

5. The water supply control method based on multiple redundant neural networks according to claim 4 is characterized in that: Compare the obtained output proportional coefficient with the average output proportional coefficient of the supercharging equipment in the zoned area. >85%, no response will be made. When ≤85%, a pressure regulation command is issued, and the output proportional coefficients of the two boosting devices before and after the boosting device in the water supply system are obtained, which are recorded as and ,like ≤85% or ≤85%, a maintenance order will be issued to repair the voltage stabilizing equipment; if >85% or >85%, the voltage regulation instruction is executed to adjust the and Output power of the corresponding booster equipment; Among them, when the booster equipment is adjusted, As the adjustment ratio of output power.

6. The water supply control method based on multiple redundant neural networks according to claim 5 is characterized in that: The decision fusion algorithm includes but is not limited to the weighted average algorithm and the voting algorithm. The final water supply control plan is determined by fusing the output results of multiple neural network models. After confirming the optimal plan, it is achieved by comparing the output results of multiple neural network models and comparing the output results with the preset normal range to determine whether there is a fault in the model or the water supply system.

7. The water supply control method based on multiple redundant neural networks according to claim 5 is characterized in that: The analysis process of running the attenuation coefficient is as follows: Set the time interval for data retrieval, recorded as , get the current and The output proportional coefficient of the same booster equipment and the stability assessment index of the area where it is located days ago; By analyzing the formula Get running time Operational attenuation factor , represents the correction factor of the operating attenuation coefficient, where: , Respectively represent the current output proportional coefficient and Output scale factor days ago, , Respectively represent the current stability assessment index of the area where the same booster equipment is located and Stability assessment index days ago; If the operating attenuation factor When the preset balance threshold is exceeded, a maintenance check instruction is issued to the maintenance department, and the operation attenuation coefficient is recalculated based on the data obtained from the actual test, and the correction coefficient is adjusted based on the difference between the two calculations; Otherwise, no response will be given.

8. The water supply control method based on multiple redundant neural networks according to claim 7 is characterized in that: The two calculated running attenuation coefficients are compared. When the difference between the two is within the preset error tolerance threshold range, a qualified instruction is issued and no correction coefficient is made. When the difference between the two is greater than the upper limit threshold in the error tolerance threshold range, an upward adjustment instruction is issued; otherwise, a downward adjustment instruction is issued; When receiving an upward command, as a correction factor for the corrected operational attenuation factor; When receiving a downward adjustment instruction, As the correction factor for the corrected operating attenuation factor, ∈(0.12, 0.15).

9. The water supply control method based on multiple redundant neural networks according to claim 7 is characterized in that: The formula for calculating the state estimation coefficient is as follows: ； In the formula, represents the state estimation coefficient of the kth booster equipment in the divided area, represents the operating attenuation coefficient of the i-th booster equipment, , They represent the maximum and minimum operating attenuation coefficients of the booster equipment in the divided area, respectively. They represent the weights of the preset operation attenuation coefficient and output proportional coefficient respectively. ; when When ≤1, Recorded as 1; when > When , the state estimation coefficient obtained in this calculation is discarded, and the boosting equipment corresponding to the state estimation coefficient is marked.

10. The water supply control method based on multiple redundant neural networks according to claim 9, characterized in that: When the state prediction coefficient exceeds the preset threshold, an alarm is issued, and the water supply in each period of the previous week is obtained. The period with the minimum water supply is selected as the maintenance time, and the booster equipment in the area is repaired; Otherwise, continue to monitor the current status, record relevant data, and adjust the operating parameters of the booster equipment according to the real-time water supply.

Images