Industrial circulating water cooperative control method and system based on multi-modal edge calculation

Abstract

The invention discloses an industrial circulating water cooperative control method and system based on multi-mode edge calculation. The method comprises the following steps that a vibration signal, a temperature signal and a flow signal of a circulating water system are collected through a multi-mode sensor; carrying out noise reduction, calibration and weighted fusion processing on the collected signals to generate a fusion index representing the operation state of the system; based on the fusion index and historical data, predicting a system load change trend in a future time period, and evaluating an equipment health state; if the current state or the predicted state is abnormal, starting a fault-tolerant strategy of a corresponding level, and generating a dynamic control instruction; system operation parameters are adjusted according to the execution control instruction, energy efficiency indexes are monitored in real time, and the control parameters are continuously optimized to improve system performance. According to the invention, comprehensive perception, real-time decision and efficient optimization are realized, the stability, energy efficiency and fault early warning capability of the circulating water system are improved, and energy consumption and operation and maintenance cost are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of circulating water control, in particular to an industrial circulating water collaborative control method based on multi-modal edge computing. BACKGROUND

[0002] Industrial circulating water system is the core facility to ensure equipment cooling and maintain process stability in industrial production. The advancement of its control technology is directly related to production efficiency, energy consumption and operation safety. However, the current traditional control method has many technical limitations that are difficult to overcome in practical application, mainly in the following three aspects: In terms of control strategy, the PID control algorithm with fixed parameters only relies on single sensor input signal. This simple feedback mode with single input and single output cannot adapt to complex variable load conditions, resulting in rigid control strategy and long-term deviation of the actuator from the optimal working point. The root cause lies in the backward design of the control system architecture, which lacks comprehensive analysis capability of multi-dimensional operation data.

[0003] In terms of system architecture, the cloud centralized computing mode is generally adopted, and all sensor data need to be uploaded to the remote server for processing before the control command is issued. This not only causes communication delay and cannot meet the real-time control demand, but also has single point failure risk due to centralized architecture, which is difficult to ensure system reliability.

[0004] In terms of maintenance mechanism, it completely relies on regular manual maintenance and lacks real-time equipment health status evaluation capability. This passive maintenance method cannot detect potential faults in advance, resulting in frequent unplanned shutdown, causing significant economic losses to enterprises and seriously affecting production continuity. SUMMARY

[0005] The main purpose of the present application is to provide an industrial circulating water collaborative control method and system based on multi-modal edge computing, which aims to solve the problems of one-sided data perception, rough signal processing, weak prediction and evaluation capability, single fault tolerance mechanism and lack of optimization closed loop in the prior art.

[0006] To achieve the above purpose, the present application provides the following solutions: In a first aspect, the present application provides an industrial circulating water collaborative control method based on multi-modal edge computing, comprising the following steps: S1, collecting vibration signal, temperature signal and flow signal of the circulating water system through multi-modal sensor; S2, performing noise reduction, calibration and weighted fusion processing on the collected signals to generate fusion indicators representing the system operation state; S3, predicting the system load change trend in the future period and evaluating the equipment health status based on the fusion indicators and historical data; S4. If the current state or predicted state is abnormal, activate the corresponding level of fault tolerance strategy and generate dynamic control instructions. S5. Adjust system operating parameters according to the executed control commands, monitor energy efficiency indicators in real time, and continuously optimize control parameters to improve system performance.

[0007] Optionally, in step S1, the multimodal sensor includes: A vibration sensor array is used to collect mechanical vibration signals during the operation of the water pump bearing; Infrared thermal imager is used to collect data on the temperature field distribution on the surface of heat exchanger tube bundles; Ultrasonic flow meters are installed on the main circulating water pipes and branch pipes, and calculate instantaneous flow velocity using the time-difference method.

[0008] Optionally, the specific operation steps of step S2 include: The vibration signals acquired by the vibration sensor array are denoised using a wavelet transform algorithm. The flow signal is filtered using an exponentially weighted moving average. Based on the temperature field distribution obtained by the infrared thermal imager, a spatial interpolation algorithm is used to correct the single-point measurement deviation. The processed vibration signal Temperature signal Flow signal Normalize each signal to the [0, 1] interval to obtain the standardized signal. ; Weight coefficients were obtained through training based on historical system operation data and failure cases. ,in The value range is 0.3-0.5. The value range is 0.2-0.4. The value range is 0.2-0.4, and it satisfies... ; Through formula Calculate the fusion index This indicator is used to comprehensively characterize the operating status of a circulating water system.

[0009] Optionally, in step S3, the specific operation of predicting the system load change trend in future periods based on fused indicators and historical data includes: Extract the time series of the fusion indicator over the past 24 hours, denoted as ; Input the above time series data into the trained LSTM neural network model to output the load forecast for the next 2 hours. The prediction formula is:

[0010] LSTM is a Long Short-Term Memory neural network model with an input layer dimension of 24, a hidden layer containing 128 neurons, and an output layer dimension of 1. The model is trained using the mean squared error loss function.

[0011] In the formula, This represents the actual load value at time t+2. When the predicted value deviates from the historical load for the same period by more than 15%, the model parameter correction is initiated. The output layer weights are retrained using three newly added sets of recent historical data to ensure that the prediction error is ≤8%.

[0012] Optionally, in step S3, the specific operations for assessing the health status of the device include: Based on 30 consecutive days of normal system operation data, the baseline values ​​of each sensor signal were calculated. and normal fluctuation range ,in These correspond to vibration signal, temperature signal, and flow signal, respectively. Device health is calculated using a multi-parameter weighted algorithm. The formula is:

[0013] In the formula: for Time of the first Measured values ​​of similar sensors; This represents the total number of sensors; when A value ≥0.9 is considered "healthy". When 0.7≤ When the value is less than 0.9, it is considered a "state of attention"; when When the value is less than 0.7, it is determined to be in "early warning state" and fault tracing analysis is triggered.

[0014] Optionally, the assessment of device health status also includes a cross-validation mechanism: When vibration signal Exceeding the benchmark value by 30%, that is ≥1.3 And simultaneously satisfy the temperature signal ≥ At +0.5℃, even If the score is ≥0.7, the health score will still be lowered through a modified formula:

[0015] In the formula, To ensure the corrected health status, mechanical wear-related faults are prioritized for identification.

[0016] Optionally, in step S4, the specific methods for determining that the current state or predicted state is abnormal include: Based on historical normal operating condition data, a normal threshold range for the fusion indicators is set. ,in, The lower threshold is... The upper limit threshold; If the current fusion index If the predicted value for the next 2 hours is determined to be abnormal, it is considered an abnormal current state. ,or If the deviation from the historical average for the same period is ≥20%, it is judged as an abnormal prediction status. The abnormality level is quantified based on the degree of deviation, using the following formula:

[0017] or

[0018] In the formula, For system target operating indicators; when <10% is considered mildly abnormal, 10% ≤ A percentage less than 30% is considered moderately abnormal. A score of ≥30% indicates a severe abnormality.

[0019] Optionally, in step S4, activating the corresponding level of fault tolerance strategy includes: Minor anomaly: Initiate primary fault tolerance and dynamically adjust equipment operating parameters using the following formula:

[0020] In the formula, This refers to the pump speed adjustment amount. The current rotational speed, This represents the fan angle adjustment amount; the sign is determined by the direction of the deviation. Moderate anomaly: Initiate intermediate fault tolerance, switch to standby equipment and allocate load. The standby pump unit activation ratio is:

[0021] In the formula, This refers to the percentage of load borne by the standby pump, i.e., the load borne by the standby pump. The load of the original operating pump has been reduced to (times the current load). ) times; Severe anomaly: Activate advanced fault tolerance, refactor the process route, shut down the faulty branch, and distribute traffic to other branches with the following distribution factor:

[0022] In the formula, For the first New traffic flow on a normal branch, For its original traffic, For the flow rate of the faulty branch, This represents the sum of the original flows of all normal branches.

[0023] Optionally, in step S4, the formula for generating the dynamic control command is:

[0024] In the formula, This represents the current deviation. For dynamic PID parameters, satisfy:

[0025]

[0026]

[0027] Control commands When applied to the actuator, it needs to be converted into a physical quantity: pump speed. Fan angle .

[0028] Secondly, this application provides an industrial circulating water collaborative control system based on multimodal edge computing, comprising: A multimodal sensing layer is used to collect multi-dimensional operational data of industrial circulating water systems, including vibration signals, temperature signals, and flow signals. The edge computing layer, which is communicatively connected to the multimodal sensing layer, is used to perform localized processing, state assessment, load prediction, and fault-tolerant decision-making on the collected multi-dimensional data, and to generate control commands. The execution control layer is communicatively connected to the edge computing layer and is used to receive and execute the control commands to adjust the operating parameters of the circulating water system. An energy efficiency feedback layer is used to monitor system energy efficiency indicators in real time and feed the energy efficiency data back to the edge computing layer; The cloud collaboration layer communicates with the edge computing layer and is used to store historical data and provide global monitoring and remote management functions.

[0029] Through the above technical solutions, the beneficial effects of this invention are as follows: This application's industrial circulating water collaborative control method and system based on multimodal edge computing constructs a multimodal sensor network using a vibration sensor array, an infrared thermal imager, and an ultrasonic flow meter, breaking through the limitations of single data and achieving comprehensive perception of mechanical, temperature, and flow status; leveraging the localized processing of the edge computing layer, combined with wavelet denoising, exponential weighted filtering, spatial interpolation calibration, and weighted fusion algorithms, it improves signal quality and state characterization accuracy, solving the pain point of coarse traditional processing; utilizing LSTM neural networks to predict load trends, multi-parameter weighted algorithms to assess equipment health, and introducing a cross-validation mechanism, it achieves forward-looking regulation and accurate fault warning, compensating for insufficient prediction and evaluation capabilities; designing a three-level fault-tolerant strategy and a dynamic PID instruction generation mechanism, it adopts differentiated measures for different anomaly levels, enhancing the system's anti-interference capability; through energy efficiency feedback and gradient descent method, it continuously optimizes control parameters, significantly improving energy efficiency and operational stability, ultimately achieving comprehensive benefits of reduced energy consumption, fewer faults, and continuous production. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention, and the embodiments in the accompanying drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 A schematic flowchart of an industrial circulating water collaborative control method based on multimodal edge computing provided in an embodiment of this application; Figure 2 A schematic diagram of the structure of an industrial circulating water collaborative control system based on multimodal edge computing provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a computer device provided in one embodiment of this application; The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0032] It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of the application. Rather, these embodiments are provided to make the disclosure more thorough and complete, and to fully convey the scope of the disclosure to those skilled in the art.

[0033] The foregoing and other technical contents, features and effects of the present invention are described in conjunction with the appendix below. Figures 1-3The detailed description of the embodiments will make this clear. All structural details mentioned in the following embodiments are based on the accompanying drawings.

[0034] Exemplary embodiments of the present invention will now be described with reference to the accompanying drawings.

[0035] In an exemplary embodiment, as shown in Figure 1, a collaborative control method for industrial circulating water based on multimodal edge computing is provided. This method is executed by a control terminal with computing capabilities, specifically an edge computing gateway deployed at key nodes of the industrial circulating water network. This gateway possesses 4 TOPS of computing power, is equipped with a dedicated LSTM prediction chip and a decision tree fault-tolerant algorithm module, and interacts with the multimodal sensing layer, execution control layer, and cloud collaboration layer via the 5G URLLC protocol. It can independently complete data processing, decision generation, and command issuance, while also supporting collaborative management with a cloud platform. In this embodiment, the method includes the following steps S1 to S5: Step S1: Collect vibration signals, temperature signals and flow signals of the circulating water system through multimodal sensors.

[0036] Specifically, the multimodal sensors include a vibration sensor array, an infrared thermal imager, and an ultrasonic flow meter. The vibration sensor array is deployed on the pump bearing housing to capture the mechanical vibration signals generated by the pump bearing during operation. These signals directly reflect the bearing's wear level, assembly accuracy, and operational stability. For example, when early wear occurs in the bearing, the high-frequency components of the vibration signal will significantly increase, providing initial evidence for fault early warning.

[0037] An infrared thermal imager is installed on the surface of the heat exchanger tube bundle and collects temperature field distribution data through a non-contact measurement method. It can clearly show the temperature differences in different areas of the tube bundle, which can not only intuitively reflect the uniformity of heat exchange efficiency, but also detect local overheating caused by scaling, blockage and other problems in a timely manner, providing spatial temperature characteristics for evaluating heat exchanger performance.

[0038] Ultrasonic flow meters are installed in the main pipe and branch pipes of the circulating water system. They calculate instantaneous flow velocity using the time-of-flight method, which measures the time difference between the propagation of ultrasonic waves in the upstream and downstream directions and combines this with pipe parameters to calculate the real-time flow rate. This measurement method is adaptable to different pipe diameters and flow velocity ranges. It can monitor both the macroscopic changes in the total system flow and the flow distribution in each branch pipe, providing data support for determining the balance of the pipe network.

[0039] Step S2 involves denoising, calibrating, and weighted fusion processing of the acquired signals to generate fusion indicators characterizing the system's operating status. The specific steps are as follows: First, wavelet transform algorithm is used to denoise the signals acquired by the vibration sensor array. Specifically, the db5 wavelet basis function is selected to decompose the signal into four levels. The high-frequency noise coefficients obtained from the decomposition are filtered out using hard thresholding, retaining the signal components in the 1kHz-10kHz frequency band. This frequency band contains key characteristics of mechanical faults such as water pump bearing wear and component loosening. Noise reduction prevents environmental interference from masking the fault signals, ensuring that subsequent analysis can accurately capture subtle changes in the equipment's mechanical state.

[0040] Secondly, the flow signal acquired by the ultrasonic flow meter is filtered using an exponentially weighted moving average, with the smoothing coefficient set to 0.2-0.3. This processing effectively eliminates measurement fluctuations caused by water flow pulsations in the pipeline, making the flow data more stably reflect the actual flow state and providing a reliable basis for judging the balance of the pipeline network and load changes.

[0041] Meanwhile, based on the temperature field distribution data acquired by the infrared thermal imager, a spatial interpolation algorithm is used to correct the single-point measurement deviation of the contact temperature sensor. Since differences in sensor installation location may lead to inaccurate local temperature measurements, correction is made through the spatial correlation of the temperature field, ensuring that temperature data from different locations are spatially consistent and improving the accuracy of heat exchange status assessment.

[0042] After signal preprocessing, the processed vibration signal Temperature signal Flow signal Normalize each signal to the [0, 1] interval to obtain the standardized signal. ; Weight coefficients were obtained through training based on historical system operation data and failure cases. ,in The value range is 0.3-0.5. The value range is 0.2-0.4. The value range is 0.2-0.4, and it satisfies... ; Through formula Calculate the fusion index This indicator is used to comprehensively characterize the operating status of a circulating water system. Among them, , , This is obtained through training with historical data, for example, when the system is more susceptible to mechanical failures. The value of will be relatively increased, giving the vibration signal a higher weight in the fusion index; however, under high temperature and high load conditions, The value of can be increased to enhance the characterization of the temperature signal on the heat exchange state of the system.

[0043] Step S3: Based on the fusion indicators and historical data, predict the system load change trend in the future period and assess the health status of the equipment.

[0044] Specifically, based on fusion indicators and historical data, the specific operations for predicting future system load change trends include: Extract the time series of the fusion indicator over the past 24 hours, denoted as ; Input the above time series data into the trained LSTM neural network model to output the load forecast for the next 2 hours. The prediction formula is:

[0045] LSTM is a Long Short-Term Memory neural network model with an input layer dimension of 24 (corresponding to 24 hours of historical data), a hidden layer containing 128 neurons, and an output layer dimension of 1 (corresponding to the predicted value for the next 2 hours). The model is trained using the mean squared error loss function.

[0046] In the formula, This represents the actual load value at time t+2. When the predicted value deviates from the historical load for the same period by more than 15%, the model parameter correction is initiated. The output layer weights are retrained using three newly added sets of recent historical data to ensure that the prediction error is ≤8%.

[0047] The specific procedures for assessing the health status of equipment, as described above, include: Based on 30 consecutive days of normal system operation data, the baseline values ​​of each sensor signal were calculated. and normal fluctuation range ,in These correspond to vibration signal, temperature signal, and flow signal, respectively. Device health is calculated using a multi-parameter weighted algorithm. The formula is:

[0048] In the formula: for Time of the first Measured values ​​of the sensor (vibration signal unit is mm / s, temperature signal unit is ℃, flow rate signal unit is m³ / h); This represents the total number of sensors; when A value ≥0.9 is considered "healthy". When 0.7≤ When the value is less than 0.9, it is considered a "state of attention"; when When the value is less than 0.7, it is determined to be in "early warning state" and fault tracing analysis is triggered.

[0049] Furthermore, assessing equipment health also includes a cross-validation mechanism: When vibration signal Exceeding the benchmark value by 30%, that is ≥1.3 And simultaneously satisfy the temperature signal ≥ At +0.5℃, even If the score is ≥0.7, the health score will still be lowered through a modified formula:

[0050] In the formula, To ensure the corrected health status, mechanical wear-related faults are prioritized for identification.

[0051] Load forecasting uses an LSTM model to anticipate load change trends for the next two hours, enabling the system to proactively adjust operating parameters to adapt to load fluctuations and avoid energy efficiency losses caused by passive adjustments. Equipment health status assessment uses benchmark comparison, health quantification, and cross-validation to achieve real-time monitoring and fault warning of equipment status, identify potential problems in advance, and reduce the risk of unplanned downtime.

[0052] Step S4: If the current state or the predicted state is abnormal, activate the corresponding level of fault tolerance strategy and generate dynamic control instructions. In step S4, the specific methods for determining that the current state or predicted state is abnormal include: Based on historical normal operating condition data, a normal threshold range for the fusion indicators is set. ,in, The lower threshold is... The upper limit threshold; If the current fusion index If the predicted value for the next 2 hours is determined to be abnormal, it is considered an abnormal current state. ,or If the deviation from the historical average for the same period is ≥20%, it is judged as an abnormal prediction status. To further quantify the degree of anomaly, the system calculates the anomaly level using a deviation formula: For the current state, the deviation formula is: ; For the predicted state, the deviation formula is:

[0053] In the formula, These are the target performance indicators for the system.

[0054] Based on the calculation results, the anomaly level is divided into three levels: when <10% is considered mildly abnormal, 10% ≤ A percentage less than 30% is considered moderately abnormal. A score of ≥30% indicates a severe abnormality.

[0055] In step S4, the system activates the corresponding fault tolerance strategy for different levels of abnormal states, as follows: Minor anomaly: Initiate primary fault tolerance and dynamically adjust equipment operating parameters using the following formula:

[0056] In the formula, This refers to the pump speed adjustment amount. The current rotational speed, This is the fan angle adjustment amount, and the sign is determined by the direction of the deviation. For example, a positive value is used when the load is too high to increase the output, and a negative value is used when the load is too low to reduce the output. By fine-tuning the equipment parameters, small deviations can be quickly compensated to prevent abnormal expansion.

[0057] Moderate anomaly: Initiate intermediate fault tolerance, switch to standby equipment and allocate load. The standby pump unit activation ratio is:

[0058] In the formula, This refers to the percentage of load borne by the standby pump, i.e., the load borne by the standby pump. The load of the original operating pump has been reduced to (times the current load). This strategy distributes load pressure through equipment redundancy, maintaining the total system output capacity when the main equipment performance degrades, thus buying time for troubleshooting.

[0059] Severe anomaly: Activate advanced fault tolerance, refactor the process route, shut down the faulty branch, and distribute traffic to other branches with the following distribution factor:

[0060] In the formula, For the first New traffic flow on a normal branch, For its original traffic, For the flow rate of the faulty branch, This represents the sum of the original flow rates of all normal branches. By redistributing the flow, we ensure that the system as a whole can still meet the process cooling requirements even if a local branch fails, minimizing production interruption losses.

[0061] The hierarchical implementation of the above three-level fault tolerance strategy not only ensures efficient fine-tuning during minor anomalies, but also achieves safe redundancy and process reconfiguration during moderate to severe anomalies, forming a hierarchical fault response system that significantly improves the system's anti-interference capability and operational reliability.

[0062] In step S4, the dynamic control command is generated based on the proportional-integral-derivative (PID) control algorithm, and the specific formula is as follows:

[0063] In the formula, This represents the current deviation. To adapt to the dynamic adjustment requirements of different anomaly levels, PID parameters A dynamic adjustment mechanism is adopted to satisfy:

[0064]

[0065]

[0066] Control commands When applied to the actuator, it needs to be converted into a physical quantity: pump speed. Fan angle This conversion mechanism directly links abstract control commands with the physical parameters that the equipment can execute. This ensures that the adjustment accuracy matches the system response speed, and also adapts to the adjustment characteristics of different equipment through differentiated conversion coefficients, ensuring that all execution actions are coordinated and consistent, and achieving precise correction of the system state.

[0067] Step S5 involves adjusting system operating parameters according to the executed control commands, monitoring energy efficiency indicators in real time, and continuously optimizing control parameters to improve system performance. The specific workflow is as follows: The first step is to adjust the operating parameters, and the system adjusts them according to the control commands. The core equipment parameters were precisely adjusted, specifically as follows: the water pump speed was adjusted according to... Adjustment, among which, The reference speed before adjustment (unit: r / min). The speed adjustment coefficient (5-10 r / min / unit command) ensures a linear speed response to commands; the fan angle is adjusted accordingly. Adjustment, Reference angle (unit: °). An angle adjustment coefficient (0.5-1° / unit command) enables precise control of the blade angle; the opening of the intelligent regulating valve is adjusted according to... change, Reference opening (unit: %) The opening adjustment coefficient (0.2-0.5% / unit instruction) ensures the accuracy of flow distribution.

[0068] Secondly, energy efficiency indicators are monitored. The system calculates two key indicators in real time: one is the system performance coefficient. ,in, Cooling capacity of circulating water (unit: kW). The first is the total system input power (unit: kW), which directly reflects energy utilization efficiency; the second is pipeline pressure loss. This involves the pipeline friction coefficient. Pipe section length (Unit: m) Inner diameter (Unit: m), water density (Unit: kg / m³) and flow velocity (Unit: m / s), used to assess the operating status of the pipeline network.

[0069] Finally, the control parameters are optimized to... With the goal of maximizing, the gradient descent method is used to iteratively update the PID parameters: ,in, The learning rate is 0.01-0.05. When there are 3 consecutive iterations... If the fluctuation is ≤3%, stop this round of optimization.

[0070] By adjusting the above parameters, control commands are translated into actual equipment actions, ensuring that the system operates according to the target state; energy efficiency monitoring provides real-time insight into operating efficiency and pipeline status, providing a basis for optimization; and continuous optimization of PID parameters ensures that the system maintains high-efficiency operation under varying operating conditions, ultimately achieving closed-loop control that reduces energy consumption and improves performance.

[0071] Based on the same inventive concept, this application also provides an industrial circulating water collaborative control system based on multimodal edge computing for implementing the aforementioned industrial circulating water collaborative control method based on multimodal edge computing. The solution provided by this system is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more embodiments of the industrial circulating water collaborative control system based on multimodal edge computing provided below can be found in the limitations of the industrial circulating water collaborative control method based on multimodal edge computing described above, and will not be repeated here.

[0072] In one exemplary embodiment, such as Figure 2 As shown, an industrial circulating water collaborative control system based on multimodal edge computing is provided. This system includes a multimodal sensing layer, an edge computing layer, an execution control layer, an energy efficiency feedback layer, and a cloud collaboration layer. Wherein: The multimodal sensing layer is used to collect multi-dimensional operational data of industrial circulating water systems, including vibration signals, temperature signals, and flow signals. The edge computing layer communicates with the multimodal sensing layer to perform localized processing, status assessment, load forecasting, and fault-tolerant decision-making on the collected multi-dimensional data, and to generate control commands. The execution control layer communicates with the edge computing layer to receive and execute control commands and adjust the operating parameters of the circulating water system. The energy efficiency feedback layer is used to monitor the system's energy efficiency indicators in real time and feed the energy efficiency data back to the edge computing layer; The cloud collaboration layer communicates with the edge computing layer and is used to store historical data, provide global monitoring and remote management functions.

[0073] Specifically, the cloud-based collaboration layer interconnects with the edge computing layer via a 5G network. It is responsible for storing nearly three years of historical operational data for the system, including raw sensor data, fusion indicators, control commands, and energy efficiency data. It also provides a web-based global monitoring interface, allowing administrators to remotely view system status, download operational reports, and issue remote control commands. Furthermore, the cloud-based collaboration layer periodically iterates and trains the LSTM model and health assessment algorithm of the edge computing layer. Through transfer learning, it pushes the optimized model parameters to edge nodes, achieving collaborative evolution between the edge and the cloud.

[0074] The industrial circulating water collaborative control system of this application eliminates the limitations of single data through multimodal perception, enables real-time decision-making and rapid response through edge computing, enhances the system's fault tolerance through a hierarchical fault-tolerance strategy, significantly reduces system energy consumption through energy efficiency closed-loop optimization, and expands remote management and global optimization capabilities through cloud collaboration. The collaborative operation of each layer achieves comprehensive perception and real-time response to the system's operating status, while improving decision-making efficiency and global optimization capabilities through edge computing and cloud collaboration. This effectively enhances the operational stability, energy efficiency, and fault early warning capabilities of the industrial circulating water system, significantly reduces energy consumption and maintenance costs, and ensures the continuity and economy of industrial production.

[0075] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores power grid voltage fluctuation suppression data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a collaborative control method for industrial circulating water based on multimodal edge computing.

[0076] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0077] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0078] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0079] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0080] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0081] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0082] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0083] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0084] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A collaborative control method for industrial circulating water based on multimodal edge computing, characterized in that, Includes the following steps: S1. Vibration signals, temperature signals, and flow signals of the circulating water system are collected through multimodal sensors; S2. The collected signals are denoised, calibrated, and weighted fusion processed to generate fusion indicators characterizing the system's operating status. S3. Based on fusion indicators and historical data, predict the trend of system load changes in future periods and assess the health status of equipment; S4. If the current state or predicted state is abnormal, activate the corresponding level of fault tolerance strategy and generate dynamic control instructions. S5. Adjust system operating parameters according to the executed control commands, monitor energy efficiency indicators in real time, and continuously optimize control parameters to improve system performance.

2. The method according to claim 1, characterized in that, In step S1, the multimodal sensor includes: A vibration sensor array is used to collect mechanical vibration signals during the operation of the water pump bearing; Infrared thermal imager is used to collect data on the temperature field distribution on the surface of heat exchanger tube bundles; Ultrasonic flow meters are installed on the main circulating water pipes and branch pipes, and calculate instantaneous flow velocity using the time-difference method.

3. The method according to claim 2, characterized in that, The specific steps of step S2 include: The vibration signals acquired by the vibration sensor array are denoised using a wavelet transform algorithm. The flow signal is filtered using an exponentially weighted moving average. Based on the temperature field distribution obtained by the infrared thermal imager, a spatial interpolation algorithm is used to correct the single-point measurement deviation. The processed vibration signal Temperature signal Flow signal Normalize each signal to the [0, 1] interval to obtain the standardized signal. ; Weight coefficients were obtained through training based on historical system operation data and failure cases. ,in The value range is 0.3-0.

5. The value range is 0.2-0.

4. The value ranges from 0.2 to 0.4, and satisfies... ; Through formula Calculate the fusion index This indicator is used to comprehensively characterize the operating status of a circulating water system.

4. The method according to claim 3, characterized in that, In step S3, the specific operations for predicting the system load change trend in future periods based on fused indicators and historical data include: Extract the time series of the fusion indicator over the past 24 hours, denoted as ; Input the above time series data into the trained LSTM neural network model to output the load forecast for the next 2 hours. The prediction formula is: LSTM is a Long Short-Term Memory neural network model with an input layer dimension of 24, a hidden layer containing 128 neurons, and an output layer dimension of 1. The model is trained using the mean squared error loss function. In the formula, This represents the actual load value at time t+2. When the predicted value deviates from the historical load for the same period by more than 15%, the model parameter correction is initiated. The output layer weights are retrained using three newly added sets of recent historical data to ensure that the prediction error is ≤8%.

5. The method according to claim 4, characterized in that, In step S3, the specific operations for assessing the health status of the equipment include: Based on 30 consecutive days of normal system operation data, the baseline values ​​of each sensor signal were calculated. and normal fluctuation range ,in These correspond to vibration signal, temperature signal, and flow signal, respectively. Device health is calculated using a multi-parameter weighted algorithm. The formula is: In the formula: for Time of the first Measured values ​​of similar sensors; This represents the total number of sensors; when A value ≥0.9 is considered "healthy"; When 0.7≤ When the value is less than 0.9, it is considered a "state of attention"; when When the value is less than 0.7, it is determined to be in "early warning state" and fault tracing analysis is triggered.

6. The method according to claim 5, characterized in that, The assessment of device health status also includes a cross-validation mechanism: When vibration signal Exceeding the benchmark value by 30%, that is ≥1.3 And simultaneously satisfy the temperature signal ≥ At +0.5℃, even If the score is ≥0.7, the health score will still be lowered through a modified formula: In the formula, To ensure the corrected health status, mechanical wear-related faults are prioritized for identification.

7. The method according to claim 6, characterized in that, In step S4, the specific methods for determining that the current state or predicted state is abnormal include: Based on historical normal operating condition data, a normal threshold range for the fusion indicators is set. ,in, The lower threshold is... The upper limit threshold; If the current fusion index If the predicted value for the next 2 hours is determined to be abnormal, it is considered an abnormal current state. ,or If the deviation from the historical average for the same period is ≥20%, it is judged as an abnormal prediction status. The abnormality level is quantified based on the degree of deviation, using the following formula: or In the formula, For system target operating indicators; when <10% is considered mildly abnormal, 10% ≤ A percentage less than 30% is considered moderately abnormal. A score of ≥30% indicates a severe abnormality.

8. The method according to claim 7, characterized in that, In step S4, activating the corresponding level of fault tolerance strategy includes: Minor anomaly: Initiate primary fault tolerance and dynamically adjust equipment operating parameters using the following formula: In the formula, This refers to the pump speed adjustment amount. The current rotational speed, This represents the fan angle adjustment amount; the sign is determined by the direction of the deviation. Moderate anomaly: Initiate intermediate fault tolerance, switch to standby equipment and allocate load. The standby pump unit activation ratio is: In the formula, This refers to the percentage of load borne by the standby pump, i.e., the load borne by the standby pump. The load of the original operating pump has been reduced to (times the current load). ) times; Severe anomaly: Activate advanced fault tolerance, refactor the process route, shut down the faulty branch, and distribute traffic to other branches with the following distribution factor: In the formula, For the first New traffic flow on a normal branch, For its original traffic, For the flow rate of the faulty branch, This represents the sum of the original flows of all normal branches.

9. The method according to claim 8, characterized in that, In step S4, the formula for generating the dynamic control command is: In the formula, This represents the current deviation. For dynamic PID parameters, satisfy: Control commands When applied to the actuator, it needs to be converted into a physical quantity: pump speed. Fan angle .

10. An industrial circulating water collaborative control system based on multimodal edge computing, characterized in that, include: A multimodal sensing layer is used to collect multi-dimensional operational data of industrial circulating water systems, including vibration signals, temperature signals, and flow signals. The edge computing layer, which is communicatively connected to the multimodal sensing layer, is used to perform localized processing, state assessment, load prediction, and fault-tolerant decision-making on the collected multi-dimensional data, and to generate control commands. The execution control layer is communicatively connected to the edge computing layer and is used to receive and execute the control commands to adjust the operating parameters of the circulating water system. An energy efficiency feedback layer is used to monitor system energy efficiency indicators in real time and feed the energy efficiency data back to the edge computing layer; The cloud collaboration layer communicates with the edge computing layer and is used to store historical data and provide global monitoring and remote management functions.

Images