Pump station equipment management and control early warning system based on internet of things

Abstract

The application discloses a pump station equipment management and control early warning system based on Internet of Things, and relates to the technical field of pump station equipment management and control early warning of Internet of Things. The system comprises a multi-source sensing module, which is arranged at a key node of a pump station equipment and comprises a plurality of sensors and a collecting unit; an entropy production rate calculation module, which calculates viscous dissipation, heat conduction dissipation and total entropy production rate in real time through an embedded processor; a quantized health index generation module, which generates a health index and divides states; a fault tracing module, which positions a damage source based on gradient analysis of an entropy production cloud picture; and an early warning execution module, which triggers a three-level early warning protocol when a condition is met. The system further comprises a fluid-structure coupling solver and other units, adopts edge computing, cloud scheduling and national encryption algorithm encryption, and can realize early warning and accurate tracing of hidden faults of the pump station equipment, so as to guarantee safe and efficient operation of the equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet of Things pump station equipment management and control early warning, in particular to a pump station equipment management and control early warning system based on Internet of Things. BACKGROUND

[0002] As the core facility of water conservancy projects, the health status of pump stations directly affects water supply safety and energy utilization efficiency. Ensuring the stable operation of pump station equipment is of great significance to the overall performance of the water conservancy system. During long-term operation, equipment may gradually deteriorate due to hidden faults such as impeller cavitation and bearing wear. If not detected and addressed in a timely manner, it may lead to unplanned downtime, affecting the normal operation of water conservancy projects.

[0003] The patent document "Intelligent pump station integrated control method and system" disclosed in Chinese patent CN118938744A collects pressure, flow, and temperature data through the Internet of Things and combines an expert system for threshold alarm. However, the pump station monitoring system of this technology relies on macro parameters such as pressure, flow, and temperature for threshold alarm. For example, after collecting basic data through the Internet of Things, the device state is determined by combining an expert system. However, this technology has obvious limitations: the monitoring dimension is only at the macro level, and it cannot quantify the microscopic energy irreversible processes such as fluid viscosity dissipation and heat conduction dissipation within the equipment; the early warning mechanism relies on sensor readings exceeding the threshold to trigger, making it difficult to provide early warning when the equipment is in a sub-healthy state; the fault location uses a simple mapping rule of "equipment-sensor", which cannot accurately locate the damage source when multiple equipment coupling faults occur, making it difficult to meet the needs of equipment health management and control throughout its life cycle.

[0004] The core problem of existing technology is that it only monitors changes in external input and output parameters of equipment to determine the operating state, ignoring the irreversible nature of energy transfer within the equipment and the microscopic damage accumulation process, which leads to the inability to achieve early warning of hidden equipment faults and accurate positioning of damage sources, thereby affecting the safe and stable operation of pump station equipment. SUMMARY

[0005] The purpose of the present application is to overcome the shortcomings of the prior art and provide a pump station equipment management and control early warning system based on the Internet of Things. By deploying multi-source sensing modules to collect parameters such as pressure, temperature, and vibration at key nodes of the equipment, combining an entropy production rate calculation module to quantify the microscopic energy dissipation process, using a quantumized health index generation module to evaluate the health status of the equipment, and utilizing a fault tracing module to locate the damage source, the system can achieve early warning and accurate tracing of hidden equipment faults, thereby effectively ensuring the safe and efficient operation of the equipment.

[0006] The technical solution of the present application is a pump station equipment management and control early warning system based on the Internet of Things, which includes the following modules:

[0007] The multi-source perception module is arranged at a key node of the pump station equipment, and includes a high-density pressure sensor array, a non-contact infrared temperature sensor, a three-axis vibration sensor, and a motor current harmonic acquisition unit.

[0008] The entropy production rate calculation module executes the following algorithm in real time through an embedded processor:

[0009] Viscous dissipation entropy production rate S μ is calculated as:

[0010]

[0011] where μ(T) is a temperature-dependent dynamic viscosity coefficient, which is determined by interpolation of a fluid property database, T is a temperature field, which is obtained by spatial interpolation of the infrared sensor array, u i and u j are flow velocity components in i and j directions in a three-dimensional space, respectively, which are calculated by pressure gradient inversion, x i and x j are coordinates in i and j directions in the three-dimensional space, respectively, are partial derivatives of the flow velocity component u i with respect to the coordinates x j and the partial derivatives of the flow velocity component u j with respect to the coordinates x i ;

[0012] Thermal conduction dissipation entropy production rate S k is calculated as:

[0013]

[0014] where k(T) is a temperature-dependent thermal conductivity, which is obtained by matching the equipment model from a material library, x, y, and z are rectangular coordinates in a three-dimensional space, are partial derivatives of the temperature field T in x, y, and z directions, respectively;

[0015] Total entropy production rate S total is calculated as:

[0016] S total =∫∫∫ V (S μ +S k )dV

[0017] where V is a three-dimensional space calculation domain of the pump station equipment, and ∫∫∫ V ·dV is a triple volume integral on the spatial domain V;

[0018] The quantum health index generation module executes:

[0019] where H indexS min is the entropy production rate of the device in the new state; max is the maximum allowable entropy production rate, calculated according to the fatigue limit of the material S max = 0.8 × S failurei ;

[0020] The failure source tracing module locates the damage source based on gradient analysis of the entropy production cloud map:

[0021] Failure coordinates

[0022] In the formula, is the gradient vector of the total entropy production rate S total ;

[0023] The early warning execution module triggers a three-level early warning protocol when the condition is met;

[0024] In the formula, is the time rate of change of the quantized health index, and -5% / 24h is the maximum allowable decay rate threshold of the health index within 24 hours.

[0025] Further, the entropy production rate calculation module comprises:

[0026] The fluid-structure coupling solver solves the transient flow field in real time using a domain-coupling algorithm:

[0027] The fluid domain uses an explicit characteristic line method:

[0028]

[0029] In the formula, u is the fluid velocity vector, t is the time, is the gradient operator, p1 is the fluid density, p is the fluid pressure, v is the kinematic viscosity, is the Laplace operator;

[0030] The structure domain uses an implicit generalized alpha method:

[0031]

[0032] In the formula, M0 is the structure mass matrix, C is the structure damping matrix, K is the structure stiffness matrix, u0 are the acceleration, velocity, and displacement vectors of the structure, respectively, and F is the fluid-structure coupling force vector;

[0033] The fluid-structure interface uses an iterative strong coupling algorithm:

[0034]

[0035] In the formula, u f , Let σ represent the displacements on the fluid side and the structure side at the fluid-structure interface, respectively. f σ s Let n be the stress tensor on the fluid side and the structural side at the fluid-structure interface, respectively. f n s These are the normal vectors on the fluid side and the structure side at the fluid-structure interface, respectively.

[0036] Convergence condition ||R|| < 10 -6 Where R is the interface residual vector:

[0037] Material damage accumulation model, fatigue damage degree calculation based on local entropy yield:

[0038]

[0039] In the formula, D represents the material fatigue damage degree, C,m is the damage coefficient obtained by fitting the material's SN curve, and S... local S represents the entropy production rate of a local region. ref The average value under the equipment's design operating conditions is used as the reference entropy productivity; t is time.

[0040] Dynamic mesh optimization cells, adaptive mesh refinement based on entropy yield gradient:

[0041] Grid size

[0042] In the formula, Δx0 is the initial mesh size, and α is an adjustable parameter with a default value of 0.01. Let be the magnitude of the total entropy productivity gradient.

[0043] Furthermore, the quantum health index generation module performs the following:

[0044] State discretization algorithm: Mapping continuous health indices to discrete states.

[0045]

[0046] In the formula, State represents the discrete state of the device;

[0047] Attenuation rate compensation mechanism, dynamically adjusting thresholds based on equipment service life:

[0048]

[0049] In the formula, The maximum permissible entropy production rate at service time t. t0 represents the maximum permissible entropy production rate at the initial time t0, and β is the aging coefficient.

[0050] Furthermore, the fault tracing module includes:

[0051] The entropy generation cloud map generation unit, and the three-dimensional field is constructed based on Delaunay triangulation:

[0052]

[0053] In the formula, S total (x, y, z) is the total entropy generation rate at the three-dimensional space point (x, y, z), N is the number of sensor nodes, and φ k (x, y, z) is the shape function of the kth node, S k is the entropy generation rate value of the kth sensor node;

[0054] The damage source positioning algorithm solves the extreme point of the gradient of the entropy generation rate field:

[0055]

[0056] In the formula, respectively, the partial derivative of the total entropy generation rate in the x, y, and z directions, is the gradient of the kth node shape function,

[0057] The uncertainty quantification unit calculates the positioning error radius:

[0058] In the formula, Error r is the positioning error radius, The second derivative matrix of the total entropy generation rate, is the determinant of the Hessian matrix.

[0059] Further, the multi-source perception module further comprises:

[0060] The cross-modal data fusion unit reconstructs the missing data using a physically constrained generative adversarial network:

[0061]

[0062] In the formula, G is a generator network, D is a discriminator network, is the mathematical expectation, x real is the real data, z is random noise, G(z) is the pseudo data generated by the generator, λ is the weight coefficient of the physical constraint term,

[0063] R physics is a physical rule constraint term, specifically:

[0064]

[0065] In the formula, f is the frequency, f gen is the frequency function of the generated data, f real is the frequency function of the real data, f NyqNyquist frequency, ||·|| is L2 norm;

[0066] Self-powered sensor node, vibration energy is converted into electrical energy by piezoelectric energy harvester:

[0067]

[0068] In the formula, P harvest The output power of the energy harvester, c p The energy conversion efficiency, ρ2 is the air density, A is the effective vibration area of the energy harvester, V is the vibration speed.

[0069] Further, the early warning execution module comprises:

[0070] Dual-objective optimization decision maker, solve the Pareto optimal solution of operation strategy:

[0071]

[0072] In the formula, N is the number of maintenance schemes, Risk i The risk index of the ith maintenance scheme, satisfies Cost i The cost of the ith maintenance scheme, M is the number of equipment units, P j The power consumption of the jth equipment unit;

[0073] Digital twin verification unit, pre-play disposal scheme in virtual environment:

[0074]

[0075] In the formula, Δt safe The safety response time, D(τ) is the damage diffusion rate function, The damage cumulative integral.

[0076] Further, it also includes:

[0077] Federal learning update module, when a new fault mode is detected, executes:

[0078]

[0079] In the formula, θ global The global model parameters, K is the number of pump stations, n k The fault data volume of the kth pump station, n is the total data volume of all pump stations, The local model parameters of the kth pump station at time t;

[0080] Blockchain storage unit, write early warning events into an unalterable ledger:

[0081]

[0082] In the formula, Hash is a hash value, SHA256 is an SHA-256 encryption algorithm, and Timestamp is a time stamp.

[0083] Further, the Jetson AGX Orin edge computing module is adopted to process the entropy production rate in real time, the Kubernetes cloud is adopted to dynamically schedule tasks, and the national SM9 algorithm is adopted to encrypt data transmission, specifically as follows.

[0084] The edge computing layer adopts the Jetson AGX Orin module to deploy the entropy production rate calculation.

[0085] The cloud analysis layer is based on the Kubernetes container orchestration to dynamically schedule the calculation tasks.

[0086] The secure communication protocol uses the national SM9 algorithm to encrypt the data transmission.

[0087] Ciphertext = Enc sM9 (PK BC , Plaintext)

[0088] In the formula, Ciphertext is encrypted ciphertext, Enc SM9 is an SM9 national encryption algorithm, PK B C is a blockchain public key, and Plaintext is plaintext data to be encrypted.

[0089] Compared with the prior art, the pump station equipment management and control early warning system based on the Internet of Things has the following beneficial effects:

[0090] Firstly, the present application realizes early warning and accurate tracing of hidden equipment failure through the synergistic effect of the multi-source sensing module and the entropy production rate calculation module.

[0091] Secondly, the application improves the system data processing efficiency and data transmission security by the edge computing and cloud scheduling cooperative architecture combined with the national encryption algorithm, the edge computing node is deployed in the pump station site, can process high-density sensing data in real time, and the dynamic grid optimization algorithm is used to adaptively adjust the computing resources, so that the key index analysis delay is controlled within 50ms, the cloud scheduling module uses the Kubernetes container technology, dynamically allocates computing resources according to the task load, supports parallel computing of the fluid-solid coupling solver, greatly improves the efficiency of transient flow field solution, and in the data transmission process, the SM9 national encryption algorithm is used to encrypt the sensitive data such as equipment health index and entropy rate cloud map from end to end, the key length is 256 bits, which can effectively resist man-in-the-middle attacks and data tampering, the architecture considers real-time and scalability, ensures data security, reduces dependence on on-site hardware resources, and is suitable for centralized management and control of distributed pump station groups. BRIEF DESCRIPTION OF DRAWINGS

[0092] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0093] Figure 1 It is an operation flow chart of a pump station equipment management and control early warning system based on Internet of Things;

[0094] Figure 2 It is a system composition schematic diagram of a pump station equipment management and control early warning system based on Internet of Things;

[0095] Figure 3 It is an entropy production driven health evaluation principle block diagram of a pump station equipment management and control early warning system based on Internet of Things;

[0096] Figure 4 It is a fault early warning traceability mechanism block diagram of a pump station equipment management and control early warning system based on Internet of Things. DETAILED DESCRIPTION

[0097] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined application purpose, the following will combine the drawings and the preferred embodiments to specifically describe the specific embodiments, structures, features and effects of the present application.

[0098] Embodiment one

[0099] The embodiment is based on a pump station equipment management and control early warning system based on Internet of Things, aiming to solve the problems of early detection of hidden failures and accurate positioning of damage sources in traditional pump station equipment management and control through the integration of multi-source sensing, entropy production rate calculation, quantum health assessment, fault precise tracing and intelligent early warning and other functions by Internet of Things technology. Figure 1 As shown in Figure 4, the system collects pressure, temperature, vibration and other data through multi-source sensors deployed at key nodes of the pump station, quantifies the microscopic energy dissipation process by entropy production rate, generates a quantum health index to evaluate the equipment state, locates the fault source through gradient analysis, triggers a hierarchical early warning, relies on a federated learning dynamic optimization model and a blockchain storage to ensure data credibility, and realizes intelligent management and control of the pump station equipment throughout its life cycle.

[0100] Working principle and operation process of multi-source sensing module; the multi-source sensing module is the "sensing nerve ending" of the system, responsible for collecting key physical parameters of pump station equipment operation to provide raw data support for subsequent analysis. Its deployment follows the "key node full coverage" principle, including the following units and operation processes;

[0101] Sensor array deployment; a high-density pressure sensor array is deployed in areas with significant pressure gradient changes such as pump impellers, pipe bends and valves, with 8-12 sensors per square meter, a sampling frequency of 1 kHz, and used to capture transient changes in fluid pressure; a non-contact infrared temperature sensor array covers motor housings, bearing seats and pipe outer walls, using a 32x32 pixel infrared imaging chip with a spatial resolution of 0.5mm and a sampling period of 100ms to generate a device surface temperature field distribution; a three-axis vibration sensor is installed on vibration-sensitive components such as motor stators and pump bearings, with a measurement range of 0-5000Hz and an accuracy of ±0.1g, used to collect vibration acceleration signals; a motor current harmonic acquisition unit is connected in series with the motor power supply circuit, with a sampling frequency of 20kHz, capable of capturing 1-50 harmonic components and reflecting abnormal electrical performance of the motor.

[0102] Cross-modal data fusion unit; due to environmental interference such as water vapor and electromagnetic noise, data may be missing from the sensor, so this unit uses a physically constrained generative adversarial network to reconstruct the missing data. Its core algorithm is;

[0103] minmaxE[logD(x real )]+E0log*1-D(G(z)))1+λ·R physics

[0104] Where the generator G generates pseudo data from random noise z, the discriminator D distinguishes between real data x real and pseudo data G(z), and the physical constraint term R physics ensures that the generated data conforms to physical rules. The first term guarantees the consistency of the energy spectrum of the data (f is frequency), and the second term constrains the Nyquist frequency f Nyq The signal amplitude at the sampling point is 0 (consistent with the sampling theorem). λ is the weight coefficient, which is set to 0.8 by default and is determined by historical data verification.

[0105] Self-powered sensor node; to solve the problem of sensor endurance, the node uses a piezoelectric energy harvester to convert device vibration energy into electrical energy, and the energy harvesting formula is:

[0106]

[0107] Where c p is the energy conversion efficiency, the actual conversion efficiency of the piezoelectric material is 0.35 determined by experiment, ρ2 is the air density (1.225 kg / m 3 ), A is the effective vibration area of the energy harvester (0.01 m 2 ), and V is the vibration speed (measured by a three-axis vibration sensor, unit m / s). For example, when the device vibration speed is 2 m / s, a single node can output

[0108]

[0109] Meet the low-power requirements of the sensor (standby power <0.005 W).

[0110] Working principle and operation process of entropy production rate calculation module; the entropy production rate calculation module is the "core analysis engine" of the system, which reflects the microscopic changes of the device health state by quantifying the irreversible process of energy dissipation inside the device. Its operation process includes entropy production rate calculation, fluid-structure coupling analysis, damage accumulation evaluation and grid optimization;

[0111] Viscous dissipation entropy production rate S μ ; describes the energy dissipation of fluid due to viscous friction, and the formula is:

[0112]

[0113] Where μ(T) is the temperature-dependent dynamic viscosity coefficient, which is obtained by interpolation according to temperature T through the fluid property database (such as NIST fluid database)), u i , u j are the flow velocity components in i, j directions in three-dimensional space, respectively, which are calculated by pressure gradient inversion, and the flow velocity field is obtained by pressure field inversion through high-density pressure sensors, x i , x j are the coordinates in i, j directions in three-dimensional space, are the partial derivatives of the coordinates x i to the flow velocity components u j , u jPartial derivative of coordinate x i .

[0114] Heat conduction dissipation entropy rate S k , reflecting the heat dissipation caused by temperature gradient, the formula is;

[0115]

[0116] Where k(T) is the temperature-dependent thermal conductivity (such as steel at 20℃ k≈50W / (m·K), obtained by matching the material library according to the equipment model (such as the water pump shell material is cast iron)), T is the temperature field, the discrete temperature values collected by the infrared sensor array are generated by spatial interpolation (such as Kriging interpolation method) three-dimensional temperature field distribution, x, y, z are the rectangular coordinates in three-dimensional space, , respectively, the partial derivative of the temperature field T in the x, y, z direction.

[0117] Total entropy rate S total ; integrate the above two kinds of dissipation, the formula is;

[0118] S total =∫∫∫ V (S k +S k )dV

[0119] In the formula, V is the three-dimensional space calculation domain of the pump station equipment, ∫∫∫ V ·dV is the triple volume integral of the spatial domain V;

[0120] In actual calculation, the three-dimensional space V of the equipment is discretized into 10 6 grid units, and the total entropy rate is obtained by numerical integration (such as Gaussian integration) summation, the unit is W / K.

[0121] Fluid-structure coupling solver; for accurate calculation of fluid and equipment structure interaction, divided into three parts;

[0122] Fluid domain solution; solve the transient flow field control equation by using explicit characteristic line method;

[0123]

[0124] Where u is the fluid velocity vector, t is the time, is the gradient operator, ρ1 is the fluid density, p is the fluid pressure, v is the kinematic viscosity, is the Laplace operator; v=μ(T) / ρ1 is the kinematic viscosity, the flow velocity u and pressure p are updated every time step by the explicit algorithm, which is suitable for high-speed fluid calculation.

[0125] Structure domain solution; solve the structure vibration equation by using implicit generalized α method;

[0126]

[0127] Where M0is the mass matrix, C is the structural damping matrix, K is the structural stiffness matrix (obtained by finite element analysis of the equipment CAD model), F is the fluid-structure interaction force vector (transmitted by the fluid domain), the implicit algorithm can guarantee numerical stability, and the time step is 1e-4s, u0is the acceleration, velocity, and displacement vector of the structure, respectively.

[0128] Fluid-structure interface coupling; iterative strong coupling algorithm is adopted to meet the displacement continuity (u f = u s , ufis the fluid side displacement, u s is the structure side displacement) and force balance (σ f · n f = σ s · n s , σ is the stress tensor, and n is the normal vector), and the iteration is converged when the modulus of the interface residual vector R is less than 1e-6, ensuring the calculation accuracy of fluid-structure interaction.

[0129] Material damage accumulation model; fatigue damage degree is calculated based on local entropy production rate;

[0130]

[0131] Where D is the material fatigue damage degree, S local is the local grid entropy production rate, S ref is the average entropy production rate under design conditions (such as S ref = 5 W / K at the rated flow of the water pump), C and m are damage coefficients (obtained by fitting the material S-N curve (fatigue life curve), such as C = 1e-12 and m = 3 for cast iron), the integral time, and the device operation time t is the length, D = 1 is judged as fatigue failure.

[0132] Dynamic grid optimization unit; the grid density is adaptively adjusted according to the entropy production rate gradient;

[0133]

[0134] Where Δx0= 0.1 mm is the initial grid size, α = 0.1 is the adjustable parameter, is the modulus of the entropy production rate gradient. In the area where the entropy production rate changes sharply (such as the impeller edge), the grid is automatically encrypted to Δx = 0.1 mm to improve the calculation accuracy; in the uniform area, the coarse grid is maintained to reduce the calculation amount.

[0135] Working principle and running process of the quantization health index generation module; this module converts the entropy production rate into an intuitive health index, realizes the quantitative evaluation of the equipment state, and the running process is as follows;

[0136] Health index calculation; the quantized health index formula is

[0137]

[0138] Wherein, H index is the quantized health index, S min is the entropy production rate reference value of the equipment in the new state, S max is the maximum allowable entropy production rate, calculated according to the material fatigue limit, S n ax=0.8×S failure When S total =S min , H index =100 (healthy); when S total =S max , H index =0 (failure).

[0139] State discretization algorithm; map the continuous H index to a discrete state, which is convenient for intuitive judgment;

[0140]

[0141] For example, when a water pump runs S total =10W / K, H index =100×(10-3) / (16-3)≈53.8, which is in the "warning" state.

[0142] Decay rate compensation mechanism; consider the effect of equipment aging on S max , dynamically adjust the threshold value;

[0143]

[0144] Wherein, is the initial maximum entropy production rate, t-t0 is the service life (years), and β is the aging coefficient (obtained by regression of the failure data of similar equipment in the past 10 years, such as water pump β=0.02 / year). For example, after 5 years of service, Avoid misjudgment due to equipment aging.

[0145] Working principle and operation process of fault tracing module; this module locates the damage source through entropy production rate distribution, and the operation process is as follows;

[0146] Entropy cloud map generation unit; based on Delaunay triangulation to construct three-dimensional entropy rate field, the formula is

[0147]

[0148] In the formula, Stotal (x,y,z) represents the total entropy production rate at the point (x,y,z) in three-dimensional space, φ k (x,y,z) is the shape function of the k-th node, S k Let N be the entropy productivity value of the k-th sensor node; N is the number of sensor nodes (approximately 1000). For shape functions (satisfying) ), through node entropy productivity S k Interpolation is used to generate the entropy production rate of any point (x, y, z), and finally, an entropy production cloud map is generated for visualization (the red area is the high entropy production rate area, indicating potential faults).

[0149] Damage source localization algorithm; the fault source is located at the point of maximum entropy production gradient, as shown in the formula;

[0150] Fault coordinates

[0151] in, By calculating the gradient magnitude at each point The point of maximum value is the source of damage. For example, the high gradient region of a water pump is located at the edge of the impeller blades, with coordinates (0.5m, 0.2m, 0.1m), corresponding to impeller cavitation damage.

[0152] Uncertainty quantification unit; Calculation of positioning error radius;

[0153]

[0154] in, Let be the second derivative matrix of the entropy production rate (Hessian matrix), and its determinant be... The larger the value, the steeper the gradient change and the smaller the error. hour, The positioning accuracy reaches 5mm.

[0155] The working principle and operation process of the early warning execution module: This module triggers early warnings and executes optimization decisions based on changes in health indices. The operation process is as follows:

[0156] Level 3 warning triggered; when the rate of decline of the health index meets the following conditions. Time (i.e., within 24 hours H) index A 25% decrease triggers a Level 3 alert. For example, H index The value dropped from 70 to 62 (a decrease of 11.4% within 24 hours), triggering an alert, and the system automatically pushed the alarm to the maintenance terminal.

[0157] A bi-objective optimization decision maker is used to find the Pareto optimal solution for the operating strategy, with the objective function being:

[0158]

[0159] where N is the number of maintenance schemes, Risk i is the risk index of the ith maintenance scheme, satisfying Risk factor of the ith maintenance, Cost i is the maintenance cost, P j is the device power consumption. By multi-objective genetic algorithm (NSGA-II), the "cost-energy" optimal scheme is generated, such as prioritizing maintenance of high-risk low-cost components, while adjusting the operating parameters to reduce power consumption.

[0160] Digital twin verification unit, pre-implementation of disposal scheme in virtual digital twin, calculation of safe response time;

[0161]

[0162] where D(τ) is the damage diffusion rate function, is the damage accumulation integral, τ is the integral variable, and the integral result ≥1 is judged as failure diffusion to dangerous state, Δt safe is the allowed maximum response time. For example, if the τ(t) of a certain fault is 0.1 / h, then Δt safe =10h, the maintenance needs to be completed within 10 hours.

[0163] Federal learning update module; when a new type of fault (such as unknown type of bearing wear) is detected, update the global model through federal learning;

[0164]

[0165] where θ global is the global model parameter, K is the number of pump stations, n k is the fault data volume of the kth pump station, n=∑n k , is the local model parameter of the Kth pump station, the global model θ global aggregates the data of each pump station without sharing the original data, protecting privacy.

[0166] Blockchain storage unit; write the early warning event into the blockchain, and the hash value is calculated as:

[0167]

[0168] where Timestamp is the timestamp, and SHA256 encryption ensures that the data cannot be tampered with, facilitating the tracing of the historical state of the equipment and the record of early warning and disposal.

[0169] In summary, the embodiment details the operation mechanism of the pump station equipment management and control early warning system based on the Internet of Things. Through the multi-source perception module, comprehensive collection and fusion of equipment parameters are achieved. The entropy production rate calculation module quantifies the microscopic energy dissipation process. The quantum health index generation module realizes intuitive evaluation of the equipment state. The fault tracing module accurately locates the damage source. The early warning execution module triggers a hierarchical response and optimization decision. The federal learning and blockchain technology are used to realize dynamic model updating and data credible notarization.

[0170] The system breaks through the limitations of traditional macroscopic parameter monitoring, captures hidden equipment failures through the microscopic indicator of entropy production rate, realizes early intervention through the health index decay rate early warning mechanism, and achieves millimeter-level fault tracing accuracy, significantly improving the management and control efficiency and reliability of pump station equipment.

[0171] Embodiment Two

[0172] As shown in Figure 1 , the embodiment provides a workflow of a pump station equipment management and control early warning system based on the Internet of Things. The specific steps of the process are:

[0173] This embodiment focuses on the complete workflow of the pump station equipment management and control early warning system based on the Internet of Things, from equipment operation data collection to early warning disposal closed loop, and details the execution logic and operation steps of each link of the system. The process takes "perception-analysis-evaluation-tracing-early warning-optimization" as the main line, realizes intelligent management and control of the whole life cycle of pump station equipment through modular collaboration, and intuitively presents the whole-chain operation mechanism of the system from data input to decision output.

[0174] S100, system initialization and parameter configuration

[0175] After the system deployment is completed, the initialization setting is first performed to lay the foundation for subsequent operation;

[0176] Device information input; through the man-machine interaction interface, input the basic information of the pump station equipment, including pump model, motor power, pipe material, design flow, rated pressure and other parameters, the system automatically matches the thermal conductivity, viscosity coefficient and other attribute data in the material library, and establishes the equipment digital archives;

[0177] Baseline calibration; in the brand-new state or after overhaul, start the "baseline collection mode", run continuously for 24 hours, record the original data (pressure, temperature, vibration, etc.) of the multi-source sensors, calculate and store the total entropy production rate baseline value at this time, and determine the maximum allowable entropy production rate according to the material fatigue limit as the initial threshold for health assessment;

[0178] Sensor network debugging; check the connection status of high-density pressure sensor array, infrared temperature sensor, three-axis vibration sensor and current harmonic acquisition unit, ensure the matching of sampling frequency and data transmission protocol, test data integrity through cross-modal data fusion unit, if there is missing, automatically start the reconstruction function, verify the power supply stability of self-powered sensor nodes (output voltage ≥ 3.3V).

[0179] S200, real-time data acquisition and preprocessing

[0180] After the system enters the normal operation stage, the multi-source sensing module continuously acquires data and performs preprocessing;

[0181] Multi-source data acquisition; pressure sensor array acquires pressure data of pipeline and impeller every 100 milliseconds, generating pressure field distribution map;

[0182] Infrared temperature sensor array takes a picture of the surface temperature image of the equipment every 200 milliseconds, converting it into temperature field data;

[0183] Three-axis vibration sensor acquires vibration acceleration signals of motor and bearing in real time, filtering out environmental low-frequency noise below 50Hz;

[0184] Current harmonic acquisition unit records the harmonic components of motor current every 50 milliseconds, focusing on monitoring characteristic harmonics such as 3 and 5;

[0185] Data cleaning and fusion; eliminate obvious outliers (such as pressure sudden increase by 10 times, temperature exceeding material melting point), for continuous missing data (such as temporary sensor offline), complete through cross-modal data fusion unit, ensure data timestamp synchronization (error ≤ 10 milliseconds);

[0186] Convert different types of sensor data into standardized format (such as JSON), mark equipment number, collection location and time, upload to edge computing node (Jetson AGX Orin).

[0187] S300, entropy rate calculation and health status evaluation

[0188] Edge computing node analyzes the preprocessed data and evaluates the device health status;

[0189] Entropy rate calculation; according to the pressure field data, the fluid flow rate is inverted, combined with the temperature field data to query the dynamic viscosity coefficient and thermal conductivity in the material library, calculate the viscous dissipation entropy rate and heat conduction dissipation entropy rate;

[0190] Integrate the results of the two, get the total entropy rate of the whole device through numerical integration, and generate real-time entropy rate curve;

[0191] Quantum health index generation;

[0192] Determine the health index formula according to the initialization phase;

[0193] According to the health index value, the state is divided, and different colors (green, yellow, orange, red) are displayed on the monitoring interface;

[0194] Decay rate monitoring; calculate the change rate of health index every 24 hours, if the decline amplitude ≥5%, mark as "rapid decline of health", trigger the early warning process.

[0195] S400, fault tracing and positioning

[0196] When the health index enters the "warning" or "failure" state, the fault tracing module starts the positioning process;

[0197] Entropy cloud map generation; based on the entropy production rate data of each sensor node, the entropy cloud map of the device is constructed through a three-dimensional modeling tool, and the high entropy production rate area (red) corresponds to the potential fault point, and the low entropy production rate area (blue) is the normal area.

[0198] Gradient analysis positioning; identify the area with the most dramatic change in entropy production rate (maximum gradient) in the entropy cloud map, and determine the specific location of the fault (such as the edge of the 3rd blade of the impeller, the rear bearing seat of the motor) combined with the coordinates of the three-dimensional model of the device, and calculate the positioning error radius (usually ≤5mm), marked as "high confidence fault point";

[0199] Damage type matching; compare the entropy production rate characteristics of the fault point (such as sudden temperature rise accompanied by vibration frequency 100Hz peak) with the historical fault database, match the possible damage types (such as impeller cavitation, bearing wear, motor inter-turn short circuit), and generate preliminary diagnosis results.

[0200] S500, early warning execution and decision optimization

[0201] According to the fault positioning result, the early warning execution module starts the hierarchical response;

[0202] First level warning (health index 60-80); the system displays a yellow warning on the monitoring interface and pushes a prompt message to the operator's mobile phone (such as "motor bearing sub-health, suggest strengthening monitoring");

[0203] Second level warning (health index 40-60); automatically issue sound and light alarm, generate preliminary maintenance scheme (such as "check if the impeller is cavitated, suggest to shorten the shutdown inspection period to 12 hours");

[0204] Third level warning (health index <40 or rapid decline); immediately cut off unnecessary load, start standby pump, and contact the operation and maintenance team, and push detailed fault report (including positioning coordinates, recommended maintenance tools);

[0205] Dual-target optimization decision; for secondary and above early warning, the system generates multiple treatment schemes (such as "immediate shutdown maintenance" and "reduced load operation + maintenance within 24 hours"), evaluates from cost (maintenance cost, downtime loss) and energy consumption (standby equipment power consumption), and recommends the optimal scheme;

[0206] Digital twin pre-rehearsal; simulate the execution process of the recommended scheme in the virtual digital twin system, verify whether the maintenance steps will cause secondary failure (such as whether the seal will be damaged when disassembling the impeller), and calculate the safe response time (such as "maintenance must be completed within 8 hours, otherwise the bearing may be stuck").

[0207] S600, data archiving and model updating

[0208] After the early warning treatment is completed, the system performs data archiving and model optimization;

[0209] Blockchain archiving; the key information (health index, entropy production rate gradient, fault location, treatment scheme, timestamp) of this early warning event is encrypted by SHA256 and written into the blockchain to generate an unalterable record for subsequent tracing and auditing;

[0210] Federal learning update; if this failure is a new type of failure (not matched with the historical database), the system automatically extracts its feature parameters, cooperates with the data of similar devices of other pump stations through federal learning mechanism, updates the global fault diagnosis model, and does not need to upload the original data to protect privacy;

[0211] Closed-loop evaluation; after maintenance is completed, the system re-collects device data, calculates health index recovery, evaluates treatment effect (such as "health index increases from 35 to 82 after maintenance, and the failure is eliminated"), and updates the device health record.

[0212] In summary, the workflow of the embodiment covers system initialization, data collection, health evaluation, fault tracing, early warning execution and data management, and realizes intelligent management and control of pump station equipment through modular cooperation, providing an executable execution framework for stable operation of pump station equipment.

[0213] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content without departing from the scope of the technical solution of the present application, and any equivalent embodiments with equivalent changes and modifications made on the basis of the technical essence of the present application to the above embodiments are still within the scope of the technical solution of the present application.

Claims

1. An Internet of Things-based pump station equipment management and control early warning system, characterized in that, Comprise the following modules: Multi-source perception module, deployed at key nodes of pump station equipment, comprising: high-density pressure sensor array, non-contact infrared temperature sensor, three-axis vibration sensor, and motor current harmonic acquisition unit; Entropy generation rate calculation module, which executes the following algorithm in real time through an embedded processor: Viscous dissipation entropy generation rate S μ Computed as: wherein μ(T) is a temperature-dependent dynamic viscosity coefficient, T is a temperature field, u i , j are flow velocity components in i, j directions in three-dimensional space, x i , j are coordinates in i, j directions in three-dimensional space, are partial derivatives of flow velocity components u i with respect to coordinates x j , u j with respect to coordinates x i . Heat conduction dissipation entropy production rate S k Calculated as: where k(T) is the temperature-dependent thermal conductivity, x, y, z are the rectangular coordinates in three-dimensional space, are the partial derivatives of the temperature field T in the x, y, z directions, respectively; Total entropy production rate S total Calculated as: S total = ∫∫∫ V (S μ + S k )dV where V is the three-dimensional spatial domain of the pumping station installation, and V • dV is the triple volume integral over the spatial domain V; quantization health index generation module, which performs: In the formula, H index is the quantized health index, S min is the entropy production rate reference value in the full new state of the device, S max is the maximum allowable entropy production rate; Fault tracing module, which locates the damage source based on entropy generation cloud map gradient analysis: fault coordinates wherein is the total entropy production rate S total is the gradient vector; an early warning execution module that triggers a three-stage early warning protocol when the conditions are met ​ In the formula, is the time rate of change of the quantized health index, and -5% / 24h is the maximum allowed decay rate threshold of the health index within 24 hours.

2. The pump station equipment management and early warning system based on the Internet of Things according to claim 1, characterized in that, The entropy generation rate calculation module comprises: Fluid-structure interaction solver, which uses a domain-coupled algorithm to solve the transient flow field in real time: The fluid domain uses an explicit characteristic line method: where u is the fluid velocity vector, t is time, is the gradient operator, p1is the fluid density, p is the fluid pressure, v is the kinematic viscosity, is the Laplace operator; The structure domain uses an implicit generalized alpha method: In the formula, M0 is a structural mass matrix, C is a structural damping matrix, K is a structural stiffness matrix, u0 is respectively an acceleration, a velocity, a displacement vector of the structure, and F is a fluid-structure interaction force vector. The fluid-structure interface uses an iterative strong coupling algorithm: wherein u f , are the displacements on the fluid side and on the structure side of the fluid-structure interface, respectively, σ f , σ s are the stress tensors on the fluid side and on the structure side of the fluid-structure interface, respectively, n f , n s are the normal vectors on the fluid side and on the structure side of the fluid-structure interface, respectively. The convergence condition is ∥R∥<10-6, where R is the interface residual vector: Material damage accumulation model, which calculates fatigue damage degree based on local entropy generation rate: where D is the fatigue damage of the material, C, m are damage coefficients fitted from the S-N curve of the material, S local is the entropy production rate of the local region, S ref is the reference entropy production rate, t is time; Dynamic grid optimization unit, which adaptively refines the grid according to the entropy generation rate gradient: Mesh size where Δx0is the initial grid size, and a is an adjustable parameter, is the modulus of the total entropy production rate gradient. 3.The pump station equipment management and early warning system based on the Internet of Things according to claim 1, characterized in that, The quantumized health index generation module executes: State discretization algorithm, which maps continuous health index to discrete state: In the formula, State is the discrete state of the equipment; Decay rate compensation mechanism, which dynamically adjusts the threshold according to the service life of the equipment: wherein is the maximum allowable entropy generation rate at time t, is the maximum allowable entropy generation rate at initial time t0, and β is an aging coefficient.

4. The pump station equipment management and early warning system based on the Internet of Things according to claim 1, characterized in that, The fault tracing module comprises: Entropy generation cloud map generation unit, which constructs a three-dimensional field based on Delaunay triangulation: In the formula, S total (x,y,z) is the total entropy production rate at the three-dimensional space point (x,y,z), N is the number of sensor nodes, and φ k (x,y,z) is the shape function of the kth node, S k is the entropy production rate value of the kth sensor node. Damage source positioning algorithm, which solves the extreme value point of the entropy generation rate field gradient: wherein respectively the partial derivatives of the total entropy production rate in the x, y, z directions, is the gradient of the kth nodal shape function, an uncertainty quantification unit that calculates a positioning error radius: where Error r is the positioning error radius, the second derivative matrix of the total entropy production rate, is the determinant of the Hessian matrix.

5. The pump station equipment management and early warning system based on the Internet of Things according to claim 1, characterized in that, The multi-source perception module further comprises: Cross-modal data fusion unit, which uses physical constraint generative adversarial network to reconstruct missing data: where G is a generator network, D is a discriminator network, is the mathematical expectation, x real is the real data, z is random noise, G(z) is the pseudo data generated by the generator, and λ is the weight coefficient of the physical constraint term. R physics is a physical rule constraint term, in particular: where f is the frequency, f gen is the frequency function of the generated data, f real is the frequency function of the real data, f Nyq is the Nyquist frequency, and ||·|| is the L2 norm. Self-powered sensing node, which converts vibration energy into electrical energy using a piezoelectric energy harvester: where P harvest is the output power of the energy harvester, c p is the energy conversion efficiency, p2is the air density, A is the effective area of the energy harvester, and V is the vibration velocity.

6. The pump station equipment management and early warning system based on the Internet of Things according to claim 1, characterized in that, The early warning execution module comprises: Dual-objective optimization decision maker, which solves the Pareto optimal solution of the operation strategy: where N is the number of maintenance plans, Risk i is the risk index of the i-th maintenance plan, Cost i is the cost of the i-th maintenance plan, M is the number of equipment units, P j is the power consumption of the j-th equipment unit; Digital twin verification unit, which preforms the disposal scheme in a virtual environment: where Δt safe is the safety response time, D(τ) is the damage diffusion rate function, is the damage accumulation integral.

7. The pump station equipment management and early warning system based on the Internet of Things according to claim 1, characterized in that, Also comprising: Federal learning update module, which executes when a new fault mode is detected: where θ global is the global model parameter, K is the number of pump stations, n k is the failure data volume of the kth pump station, n is the total data volume of all pump stations, is the local model parameter of the kth pump station at time t. Blockchain storage unit, which writes the early warning event into an unalterable ledger: In the formula, Hash is the hash value, SHA256 is the SHA-256 encryption algorithm, and Timestamp is the timestamp. 8.The pump station equipment management and early warning system based on Internet of Things according to any one of claims 1-7, characterized in that, Real-time processing of entropy generation rate using Jetson AGX Orin edge computing module, dynamic scheduling of tasks through Kubernetes cloud, and encryption of data transmission using national SM9 algorithm, specifically: Edge computing layer, which deploys entropy generation rate calculation using Jetson AGX Orin module; Cloud analysis layer, which dynamically schedules calculation tasks based on Kubernetes container orchestration; Secure communication protocol, which uses national SM9 algorithm to encrypt data transmission; Ciphertext = Enc SM9 (PK BC , Plaintext) In the formula, Ciphertext is the encrypted ciphertext, Enc SM9 SM9 is a national encryption algorithm, PK B is a blockchain public key, and Plaintext is the plaintext data to be encrypted.

Images