LabVIEW-based diaphragm pump test platform multi-parameter intelligent monitoring system and method

Abstract

The invention relates to the technical field of diaphragm pump testing, and discloses a LabVIEW-based diaphragm pump testing platform multi-parameter intelligent monitoring system and method, and the system comprises a hardware collection layer, an edge calculation layer, an upper computer core layer and an execution response layer. According to the method, automatic identification of medium types and adaptive calibration of compensation coefficients are completed based on preset parameters and a medium feature library; establishing a normal parameter baseline through no-load operation learning and optimizing a lightweight machine learning model; cooperatively collecting multi-dimensional parameters under a test working condition, and carrying out preprocessing and feature extraction at an edge end; intelligent fault pre-judgment is realized based on the feature data and a machine learning model; executing remote security control of encrypted transmission and hardware redundancy according to the authorization instruction; and finally, managing data through an edge-cloud collaboration mechanism and automatically generating a test report. According to the method, the calculation precision of key performance parameters and the credibility of test results are improved, and potential faults and sudden shutdown are prevented.

Description

Technical Field

[0001] This invention relates to the field of diaphragm pump testing technology, specifically to a multi-parameter intelligent monitoring system and method for a diaphragm pump testing platform based on LabVIEW. Background Technology

[0002] Due to their advantages such as simple structure, good sealing performance, and wide range of media that can be transported, diaphragm pumps are widely used in key industries such as petroleum, chemical, and pharmaceutical. Their performance stability directly determines the safety of the conveying system. Therefore, comprehensive performance testing must be carried out through a testing platform during the production and R&D stages. The monitoring system is a core component of the testing platform.

[0003] Currently, diaphragm pump testing platforms based on traditional measurement and control architectures typically rely on independent sensors for parameter acquisition during operation. The acquired multi-dimensional parameter data lacks comprehensive analysis and edge preprocessing, resulting in delayed judgment of the operating status and difficulty in locating the root cause of faults, thus affecting testing efficiency and reliability.

[0004] Therefore, a multi-parameter intelligent monitoring system and method based on LabVIEW for diaphragm pump testing platform is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multi-parameter intelligent monitoring system and method for a diaphragm pump test platform based on LabVIEW. This solves the problems mentioned in the background art, such as the reliance on independent sensors for parameter acquisition during operation of diaphragm pump test platforms based on traditional measurement and control architectures, as well as the lag in judging the operating status and the difficulty in locating the root cause of faults.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multi-parameter intelligent monitoring system and method for a diaphragm pump testing platform based on LabVIEW, wherein the method includes the following steps:

[0007] S1. Perform system initialization and two-way authentication of the equipment, and load the preset parameters of the test pump type;

[0008] S2. Based on preset parameters, perform automatic identification of media type and adaptive calibration of compensation coefficient to generate accurate compensation coefficient;

[0009] S3. Control the diaphragm pump to run under no-load according to the precise compensation coefficient to collect normal operation parameters, generate the normal parameter baseline of the pump type and complete the training and optimization of the lightweight machine learning model.

[0010] S4. Based on the machine learning model, perform collaborative acquisition of multi-dimensional parameters and edge preprocessing under test conditions to generate feature data and abnormal data;

[0011] S5. Based on the feature data and the trained lightweight machine learning model, perform intelligent fault prediction, and combine the digital twin model to realize the visualization mapping and display of the operating status.

[0012] S6. Based on the operating status and authorized instructions, perform remote security control based on encrypted transmission and hardware redundancy protection;

[0013] S7. Perform edge and cloud collaborative management on the preset parameters, feature data and abnormal data, and automatically generate a test report.

[0014] Preferably, in step S1, system initialization and device two-way authentication processing are performed, and the preset parameters for loading the test pump type include the following steps:

[0015] S11. After the device is powered on, the edge computing layer and the host computer core layer complete two-way authentication through the pre-stored unique serial number of the device and the dynamic key generated in real time, and refuse unauthorized device access.

[0016] S12. After certification, the host computer core layer loads the preset parameters corresponding to the current test diaphragm pump model. The preset parameters include rated head, rated flow rate, rated voltage, rated current, rated power, rated input pressure, and rated output pressure.

[0017] Preferably, the automatic identification of media type and adaptive calibration of compensation coefficient based on preset parameters in step S2 to generate accurate compensation coefficients includes the following steps:

[0018] S21. Start the hardware acquisition layer, collect the initial pressure data and initial temperature data of the diaphragm pump at the beginning of startup through pressure and temperature sensors, and transmit them to the edge computing layer.

[0019] S22. The edge computing layer uploads the initial pressure data and initial temperature data to the host computer core layer. The host computer core layer calls its built-in medium feature library and runs the pattern matching subroutine to automatically identify the current test medium type based on the pressure fluctuation characteristics and temperature rise characteristics.

[0020] The identification is achieved by calculating the matching degree between the currently acquired feature vector and the preset medium feature vector in the feature library. To achieve this, the calculation formula is:

[0021] ;

[0022] in, The value represents the matching degree, ranging from [0, 1]. The closer the value is to 1, the higher the matching degree. The total number of feature dimensions. The number of data collected so far 3D eigenvalues For a certain preset medium template in the feature library, the first 3D eigenvalues For the first The weight coefficients of the features are pre-set based on the feature's ability to distinguish media types;

[0023] S23. Based on the identified medium type, retrieve the corresponding basic compensation coefficient from the medium feature library, and based on the initial acquisition data, perform iterative optimization calculations using the least squares method to obtain and store the accurate compensation coefficient applicable to the current test conditions.

[0024] Preferably, in step S3, controlling the diaphragm pump to run under no-load conditions to collect normal operating parameters, generating a baseline of normal parameters for the pump type, and completing the training and optimization of the lightweight machine learning model includes the following steps:

[0025] S31. Issue an unloaded operation command to control the diaphragm pump to run stably under unload conditions for a preset time;

[0026] S32. During no-load operation, multi-dimensional parameters are continuously and synchronously collected through the hardware acquisition layer. The multi-dimensional parameters include inlet and outlet pressure, pump body temperature, medium temperature and operating noise.

[0027] S33. The edge computing layer processes and analyzes the multi-dimensional parameters collected during no-load operation to generate the parameter baseline of the pump type under normal conditions. The parameter baseline includes the noise baseline, pressure fluctuation baseline and temperature baseline.

[0028] For any parameter At any point during the baseline self-learning phase instantaneous deviation The calculation formula is:

[0029] ;

[0030] in, For parameters At any moment The collected values, For parameters The average value within the collected time period. For parameters The edge computing layer continuously monitors the standard deviation within the collected time period. This ensures that the system operates within a preset range to confirm stable system operation.

[0031] S34. Simultaneously, the normal operation data collected during the idle operation period is used to train and optimize the parameters of the lightweight random forest model deployed in the edge computing layer.

[0032] Preferably, step S4, which involves collaboratively acquiring and preprocessing multi-dimensional parameters based on a machine learning model under test conditions to generate feature data and anomaly data, includes the following steps:

[0033] S41. Issue a load test command to put the diaphragm pump into test mode;

[0034] S42. Multi-dimensional parameters under test conditions are synchronously collected through the hardware acquisition layer. The multi-dimensional parameters include inlet and outlet pressure, pump body temperature, medium temperature, noise, and ambient atmospheric pressure.

[0035] S43. The adaptive signal conditioning unit integrated inside the hardware acquisition layer performs filtering, amplification, and opto-isolation processing on the acquired raw signal.

[0036] S44. The processed data is transmitted to the edge computing layer, which performs noise reduction on the data and extracts the time-domain and frequency-domain features of the data. The time-domain features include peak value and variance, and the frequency-domain features include the main frequency and harmonic amplitude, generating feature data.

[0037] S45. The edge computing layer compares the collected data with the preset threshold and the normal parameter baseline in real time, and filters out abnormal data in milliseconds.

[0038] Preferably, step S5, which involves intelligent fault prediction based on the feature data and a trained lightweight machine learning model, and combining this with a digital twin model to visualize and display the operational status, includes the following steps:

[0039] S51, The edge computing layer uploads the feature data to the host computer core layer;

[0040] S52. The host computer core layer inputs the received feature data into a trained and optimized lightweight random forest model for fault prediction, and outputs the predicted fault type and corresponding confidence level.

[0041] S53. The prediction results output by the lightweight random forest model are subjected to secondary verification using a method based on traditional rule thresholds.

[0042] S54. When the fault confidence level reaches the preset alarm threshold, the host computer core layer triggers a graded alarm, which includes a pop-up notification on the software interface and an alarm sound.

[0043] S55. At the same time, the host computer core layer runs a digital twin visualization module, which drives a virtual twin model that corresponds one-to-one with the physical diaphragm pump. The virtual twin model includes core virtual components such as the pump body, impeller, diaphragm, and inlet and outlet pipelines.

[0044] S56. The real-time collected multi-dimensional parameters and the fault prediction results are associated and mapped with the virtual twin model. When the parameters are abnormal and a fault is predicted, the corresponding virtual component is highlighted in red and flashed in the virtual twin model. It also supports viewing the historical parameter curves and fault records of any virtual component through interactive operation.

[0045] Preferably, step S6, which involves performing remote security control based on encrypted transmission and hardware redundancy protection according to the operating status and authorization instructions, includes the following steps:

[0046] S61. Users with corresponding operating permissions send start / stop control commands and emergency stop control commands downwards through the human-machine interface according to the monitoring status displayed by the host computer core layer.

[0047] S62. The host computer core layer uses an encryption algorithm to encrypt the control commands and transmits the encrypted commands to the edge computing layer through an industrial communication network.

[0048] S63. After the edge computing layer decrypts and verifies the command, it drives the solid-state relay and motor controller in the execution response layer to perform the corresponding start and stop actions.

[0049] S64. When an emergency stop control command is executed, the hardware emergency stop switch connected in series in the motor main power supply circuit will act synchronously to form hardware redundancy protection and forcibly cut off the motor power supply.

[0050] S65. The execution response layer feeds back the instruction execution results and device status to the edge computing layer in real time, and finally uploads them to the host computer core layer for status update and display.

[0051] Preferably, the system includes a hardware acquisition layer, an edge computing layer, a host computer core layer, and an execution response layer;

[0052] The hardware acquisition layer acquires multi-dimensional physical parameters during the diaphragm pump test process and conditions the acquired raw signals.

[0053] The edge computing layer is communicatively connected to the hardware acquisition layer, and performs local preprocessing, feature extraction, anomaly screening, model inference, data caching, and secure communication on the conditioned data.

[0054] The host computer core layer is built on the LabVIEW development platform and is remotely connected to the edge computing layer through an industrial communication network. It features media identification and compensation calibration, intelligent fault prediction, full-link visual monitoring, data collaborative management, and security policy control.

[0055] The execution response layer is connected to the edge computing layer, receives control commands and drives the actuator to perform actions, realizes the start-up and emergency stop control of the diaphragm pump, and has a hardware redundancy protection mechanism.

[0056] Preferably, the host computer core layer includes a media adaptive head calculation module, a lightweight machine learning fault prediction module, a digital twin visualization module, an edge-cloud collaborative data management module, and a full-link security control module;

[0057] The adaptive head calculation module has a built-in medium feature library. It automatically identifies the test medium type based on pressure and temperature characteristics, and determines the accurate compensation coefficient through iterative optimization algorithm. It calculates high-precision head by combining the collected pressure difference, medium density, gravitational acceleration, position height difference and compensation coefficient.

[0058] The lightweight machine learning fault prediction module integrates a lightweight random forest model, which predicts fault type and confidence level based on feature data uploaded from the edge computing layer, and performs secondary verification by combining traditional rule thresholds to trigger graded alarms.

[0059] The digital twin visualization module constructs and drives a virtual twin model that is one-to-one with the physical pump, enabling real-time, intuitive, and relevant visualization and interaction of operating parameters, equipment status, and fault information.

[0060] The edge-cloud collaborative data management module adopts a collaborative mechanism of edge preprocessing and cloud aggregation storage to manage the local and cloud storage, classification, and query of raw data, feature data, and abnormal data, and automatically generate structured test reports.

[0061] The full-link security control module manages the security of the entire process from device access authentication, data transmission, user operation permissions to operation log auditing.

[0062] Preferably, the hardware acquisition layer includes a multi-dimensional sensor group and an adaptive signal conditioning unit; the edge computing layer includes a microprocessor core, a local storage unit, a hardware encryption unit and a feature processing unit; and the execution response layer includes a solid-state relay, a hardware emergency stop switch and a motor controller with status feedback.

[0063] The multi-dimensional sensor group includes a pressure sensor for measuring inlet and outlet pressure, a temperature sensor for measuring pump body and medium temperature, a noise sensor for measuring operating noise, and an atmospheric pressure sensor for measuring ambient atmospheric pressure.

[0064] The adaptive signal conditioning unit integrates a filter circuit, an instrumentation amplifier, and an opto-isolator to perform anti-interference conditioning on the raw signal output by the sensor.

[0065] The microprocessor core is the main control chip of the edge computing layer, responsible for scheduling and executing data preprocessing, feature extraction, model running and communication control tasks.

[0066] The local storage unit caches the original data at a high frequency and persists the data when the network is interrupted.

[0067] The hardware encryption unit performs encryption and decryption operations on uplink data and downlink commands.

[0068] The feature processing unit extracts time-domain and frequency-domain features from the preprocessed data in real time and constructs feature vectors.

[0069] The solid-state relay and the motor controller receive instructions from the edge computing layer to control the start and stop of the diaphragm pump motor;

[0070] The hardware emergency stop switch is physically connected in series in the main power supply circuit of the diaphragm pump motor, serving as a hardware redundant safety protection device independent of software logic.

[0071] Compared with the prior art, the present invention provides a multi-parameter intelligent monitoring system and method for a diaphragm pump testing platform based on LabVIEW, which has the following beneficial effects:

[0072] 1. In this invention, the built-in medium feature library and pattern matching subroutine can automatically identify the test medium type based on the initial pressure and temperature data, and call the iterative optimization algorithm to adaptively calibrate the basic compensation coefficient to generate an accurate compensation coefficient suitable for the current working condition. This process replaces the traditional method of relying on manual experience to set the parameters, realizes high-precision head calculation with medium adaptation, solves the human setting error caused by the difference in medium properties, and improves the calculation accuracy of key performance parameters and the reliability of test results.

[0073] 2. In this invention, by performing multi-parameter collaborative acquisition and feature extraction at the edge computing layer, and combining it with a trained lightweight random forest model for fault prediction, the transformation from simple threshold over-limit alarms to early intelligent warnings based on data features is realized. At the same time, by using a digital twin model to associate and map real-time data, prediction results and virtual pump components and display them visually, maintenance personnel can intuitively and quickly locate abnormal components and view historical status, thereby improving the efficiency and accuracy of fault diagnosis and preventing potential faults and sudden shutdowns.

[0074] 3. In this invention, a full-link security control module is established, from two-way device authentication and encrypted command transmission to hardware redundancy emergency stop. At the software level, secure remote start-stop control is achieved based on authorized and encrypted communication. At the hardware level, a hardware emergency stop switch independent of software logic provides the final physical-level protection. This system is designed to prevent risks such as unauthorized access, data tampering, and control command hijacking. At the same time, it ensures that the system can still be forcibly powered off through the hardware circuit when the control system is abnormal, providing a double layer of protection for the safety of the test platform and personnel. Attached Figure Description

[0075] Figure 1 This is a schematic diagram of the architecture of the multi-parameter intelligent monitoring system for the diaphragm pump test platform based on LabVIEW according to the present invention;

[0076] Figure 2 This is a flowchart illustrating the steps of the multi-parameter intelligent monitoring method for the diaphragm pump testing platform based on LabVIEW, as described in this invention. Detailed Implementation

[0077] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] Please see Figure 1-2 The specific implementation of the multi-parameter intelligent monitoring system and method for diaphragm pump testing platform based on LabVIEW is as follows, and the method includes the following steps:

[0079] S1. Perform system initialization and two-way authentication of the equipment, and load the preset parameters of the test pump type;

[0080] S2. Based on preset parameters, perform automatic identification of media type and adaptive calibration of compensation coefficient to generate accurate compensation coefficient;

[0081] S3. Control the diaphragm pump to run under no-load according to the precise compensation coefficient to collect normal operation parameters, generate the normal parameter baseline of the pump type and complete the training and optimization of the lightweight machine learning model.

[0082] S4. Based on the machine learning model, perform collaborative acquisition of multi-dimensional parameters and edge preprocessing under test conditions to generate feature data and abnormal data;

[0083] S5. Based on feature data and a trained lightweight machine learning model, perform intelligent fault prediction and combine it with a digital twin model to realize the visualization and display of the operating status.

[0084] S6. Based on the operating status and authorized instructions, perform remote security control based on encrypted transmission and hardware redundancy protection;

[0085] S7. Perform edge and cloud collaborative management of preset parameters, feature data and abnormal data, and automatically generate test reports.

[0086] Step S1 involves system initialization and two-way device authentication, loading the preset parameters for the test pump, including the following steps:

[0087] S11. After the device is powered on, the edge computing layer and the host computer core layer complete two-way authentication through the pre-stored unique serial number of the device and the dynamic key generated in real time, and refuse unauthorized device access.

[0088] S12. After certification, the host computer core layer loads the preset parameters corresponding to the current test diaphragm pump model. The preset parameters include rated head, rated flow rate, rated voltage, rated current, rated power, rated input pressure, and rated output pressure.

[0089] Step S2 involves automatic identification of the media type and adaptive calibration of the compensation coefficient based on preset parameters to generate accurate compensation coefficients, including the following steps:

[0090] S21. Start the hardware acquisition layer, collect the initial pressure data and initial temperature data of the diaphragm pump at the beginning of startup through pressure and temperature sensors, and transmit them to the edge computing layer.

[0091] S22. The edge computing layer uploads the initial pressure data and initial temperature data to the host computer core layer. The host computer core layer calls its built-in medium characteristic library and runs the pattern matching subroutine to automatically identify the current test medium type based on the pressure fluctuation characteristics and temperature rise characteristics.

[0092] The identification is achieved by calculating the matching degree between the currently acquired feature vector and the preset medium feature vector in the feature library. To achieve this, the calculation formula is:

[0093] ;

[0094] in, The value represents the matching degree, ranging from [0, 1]. The closer the value is to 1, the higher the matching degree. The total number of feature dimensions. The number of data collected so far 3D eigenvalues For a certain preset medium template in the feature library, the first 3D eigenvalues For the first The weight coefficients of the features are pre-set based on the feature's ability to distinguish media types;

[0095] S23. Based on the identified medium type, retrieve the corresponding basic compensation coefficient from the medium feature library, and perform iterative optimization calculations using the least squares method based on the initial collected data to obtain and store the accurate compensation coefficient applicable to the current test conditions.

[0096] Step S3 involves controlling the diaphragm pump to run under no-load conditions to collect normal operating parameters, generating a baseline of normal parameters for this pump type, and completing the training and optimization of the lightweight machine learning model. Step S3 includes the following steps:

[0097] S31. Issue an unloaded operation command to control the diaphragm pump to run stably under unload conditions for a preset time;

[0098] S32. During no-load operation, multi-dimensional parameters are continuously and synchronously collected through the hardware acquisition layer. These multi-dimensional parameters include inlet and outlet pressure, pump body temperature, medium temperature, and operating noise.

[0099] S33. The edge computing layer processes and analyzes the multi-dimensional parameters collected during no-load operation to generate the parameter baseline of the pump type under normal conditions. The parameter baseline includes the noise baseline, pressure fluctuation baseline and temperature baseline.

[0100] For any parameter At any point during the baseline self-learning phase instantaneous deviation The calculation formula is:

[0101] ;

[0102] in, For parameters At any moment The collected values, For parameters The average value within the collected time period. For parameters The edge computing layer continuously monitors the standard deviation within the collected time period. This ensures that the system operates within a preset range to confirm stable system operation.

[0103] S34. At the same time, the normal operation data collected during the idle operation period is used to train and optimize the parameters of the lightweight random forest model deployed on the edge computing layer.

[0104] Step S4 involves the collaborative acquisition of multi-dimensional parameters and edge preprocessing based on a machine learning model under test conditions to generate feature data and anomaly data, including the following steps:

[0105] S41. Issue a load test command to put the diaphragm pump into test mode;

[0106] S42. Multi-dimensional parameters under test conditions are collected synchronously through the hardware acquisition layer. These multi-dimensional parameters include inlet and outlet pressure, pump body temperature, medium temperature, noise, and ambient atmospheric pressure.

[0107] S43. The adaptive signal conditioning unit integrated inside the hardware acquisition layer performs filtering, amplification, and opto-isolation processing on the acquired raw signal.

[0108] S44. The processed data is transmitted to the edge computing layer, which performs noise reduction on the data and extracts the time-domain and frequency-domain features of the data. The time-domain features include peak value and variance, and the frequency-domain features include the main frequency and harmonic amplitude, generating feature data.

[0109] Variance in time-domain features The calculation formula is:

[0110] ;

[0111] in, The variance of a parameter sequence within a single calculation window. To calculate the number of sampling points within the window, For this parameter in the first The value of each sampling point, This is the average value of the parameter within the calculation window;

[0112] S45. The edge computing layer compares the collected data with preset thresholds and normal parameter baselines in real time, and filters out abnormal data in milliseconds.

[0113] Step S5 involves intelligent fault prediction based on feature data and a pre-trained lightweight machine learning model, combined with a digital twin model to visualize and display the operational status. This includes the following steps:

[0114] S51, the edge computing layer uploads feature data to the host computer core layer;

[0115] S52. The host computer core layer inputs the received feature data into the trained and optimized lightweight random forest model for fault prediction, and outputs the predicted fault type and corresponding confidence level.

[0116] Confidence The calculation is based on the voting results of all decision trees within the model for the current sample, and the formula is:

[0117] ;

[0118] in, The fault prediction confidence level output by the model has a value range of [0, 1]. The number of decision trees that vote for the current predicted fault type. The total number of decision trees in the lightweight random forest model;

[0119] S53. The prediction results output by the lightweight random forest model are subjected to secondary verification using a method based on traditional rule thresholds.

[0120] S54. When the fault confidence level reaches the preset alarm threshold, the host computer core layer triggers a graded alarm, which includes a pop-up notification on the software interface and an alarm sound.

[0121] S55. At the same time, the host computer core layer runs a digital twin visualization module. The digital twin visualization module drives a virtual twin model that corresponds one-to-one with the physical diaphragm pump. The virtual twin model includes the core virtual components of the pump body, impeller, diaphragm and inlet and outlet pipelines.

[0122] S56. The real-time collected multi-dimensional parameters and fault prediction results are associated and mapped with the virtual twin model. When the parameters are abnormal or a fault is predicted, the corresponding virtual component is highlighted in red and flashed in the virtual twin model. It also supports viewing the historical parameter curves and fault records of any virtual component through interactive operation.

[0123] Step S6, based on the operating status and authorized instructions, executes remote security control based on encrypted transmission and hardware redundancy protection, including the following steps:

[0124] S61. Users with the corresponding operating permissions can send start / stop control commands and emergency stop control commands to the user through the human-machine interface based on the monitoring status displayed by the host computer core layer.

[0125] S62. The host computer core layer uses an encryption algorithm to encrypt control commands and transmits the encrypted commands to the edge computing layer through an industrial communication network.

[0126] To ensure instruction integrity, the host computer core layer appends an instruction verification code when sending encrypted instructions. The calculation formula is as follows:

[0127] ;

[0128] in, The calculated single-byte instruction checksum, This represents the total number of bytes in the encrypted instruction data packet. For the first in the data packet The value of 1 byte, For bitwise XOR operation, For modulo operation;

[0129] S63. After the edge computing layer decrypts and verifies the command, it drives the solid-state relay and motor controller in the execution response layer to perform the corresponding start and stop actions.

[0130] S64. When an emergency stop control command is executed, the hardware emergency stop switch connected in series in the motor main power supply circuit will act synchronously to form hardware redundancy protection and forcibly cut off the motor power supply.

[0131] S65. The execution response layer feeds back the command execution results and device status to the edge computing layer in real time, and finally uploads them to the host computer core layer for status update and display.

[0132] The system comprises a hardware acquisition layer, an edge computing layer, a host computer core layer, and an execution response layer;

[0133] The hardware acquisition layer collects multi-dimensional physical parameters during the diaphragm pump test process and conditions the acquired raw signals.

[0134] The edge computing layer communicates with the hardware acquisition layer to perform local preprocessing, feature extraction, anomaly screening, model inference, data caching, and secure communication on the conditioned data.

[0135] The host computer core layer is built on the LabVIEW development platform and is remotely connected to the edge computing layer through an industrial communication network. It features media identification and compensation calibration, intelligent fault prediction, full-link visual monitoring, data collaborative management, and security policy control.

[0136] The execution response layer connects to the edge computing layer, receives control commands and drives the actuator to achieve the start-up, shutdown and emergency stop control of the diaphragm pump, and has a hardware redundancy protection mechanism.

[0137] The core layer of the host computer includes a media adaptive head calculation module, a lightweight machine learning fault prediction module, a digital twin visualization module, an edge-cloud collaborative data management module, and a full-link security control module;

[0138] The media-adaptive head calculation module has a built-in media feature library. It automatically identifies the test media type based on pressure and temperature characteristics, and determines the accurate compensation coefficient through iterative optimization algorithms. Combining the collected pressure difference, media density, gravitational acceleration, position height difference, and compensation coefficient, it calculates a high-precision head. Its accounting formula is:

[0139] ;

[0140] in, To calculate the head, This is the measured value of the outlet pressure. This is the measured value of the inlet pressure. For the density of the medium, It is the acceleration due to gravity. The vertical height difference between the inlet and outlet pressure measurement points. These are the accurate compensation coefficients after adaptive calibration. To match the viscosity of the medium and temperature Related characteristic functions;

[0141] The lightweight machine learning fault prediction module integrates a lightweight random forest model. It predicts fault types and confidence levels based on feature data uploaded from the edge computing layer, and performs secondary verification by combining traditional rule thresholds to trigger tiered alarms.

[0142] The digital twin visualization module constructs and drives a virtual twin model that is one-to-one with the physical pump, enabling real-time, intuitive, and relevant visualization and interaction of operating parameters, equipment status, and fault information.

[0143] The edge-cloud collaborative data management module adopts a collaborative mechanism of edge preprocessing and cloud aggregation storage to manage the local and cloud storage, classification, and query of raw data, feature data, and abnormal data, and automatically generate structured test reports.

[0144] The end-to-end security control module manages the security of the entire process, from device access authentication, data transmission, user operation permissions to operation log auditing.

[0145] The hardware acquisition layer includes a multi-dimensional sensor group and an adaptive signal conditioning unit; the edge computing layer includes a microprocessor core, a local storage unit, a hardware encryption unit, and a feature processing unit; and the execution response layer includes solid-state relays, hardware emergency stop switches, and a motor controller with status feedback.

[0146] The multi-dimensional sensor group includes a pressure sensor for measuring inlet and outlet pressure, a temperature sensor for measuring pump body and medium temperature, a noise sensor for measuring operating noise, and an atmospheric pressure sensor for measuring ambient atmospheric pressure.

[0147] The adaptive signal conditioning unit integrates a filter circuit, an instrumentation amplifier, and an opto-isolator. It performs anti-interference conditioning on the raw signal output from the sensor, conducts a preliminary assessment of the quality of the conditioned signal, and calculates the effective value of the signal. :

[0148] ;

[0149] in, For the effective value of the signal voltage within a single evaluation window, The number of sampling points within the evaluation window. For the first The instantaneous voltage value at each sampling point;

[0150] The microprocessor core is the main control chip for the edge computing layer, responsible for scheduling and executing data preprocessing, feature extraction, model running, and communication control tasks.

[0151] Local storage units cache raw data at high frequency and persist data storage in the event of network interruption;

[0152] The hardware encryption unit performs encryption and decryption operations on uplink data and downlink commands to ensure data transmission security.

[0153] The feature processing unit extracts time-domain and frequency-domain features from the preprocessed data in real time and constructs feature vectors;

[0154] Solid-state relays and motor controllers receive instructions from the edge computing layer to control the start and stop of the diaphragm pump motor;

[0155] The hardware emergency stop switch is physically connected in series in the main power supply circuit of the diaphragm pump motor, serving as a hardware redundant safety protection device independent of software logic.

[0156] The operation steps of the multi-parameter intelligent monitoring system and method for diaphragm pump testing platform based on LabVIEW are as follows:

[0157] Step 1: System Initialization and Two-Way Device Authentication

[0158] After the system is powered on, the initialization process is executed first. The edge computing layer and the host computer core layer will complete two-way identity authentication through the pre-stored unique serial number of the device and the real-time generated dynamic key to reject the access of any unauthorized device and ensure the security of the underlying system. After successful authentication, the host computer core layer will load all the preset parameters corresponding to the diaphragm pump model under test. These parameters constitute the test benchmark and usually include rated head, rated flow, rated voltage, rated current, rated power, rated input pressure and rated output pressure.

[0159] Step Two: Automatic Media Type Identification and Adaptive Calibration of Compensation Coefficients

[0160] The system initiates the hardware acquisition layer, collecting initial pressure and temperature data during the initial startup of the diaphragm pump using pressure and temperature sensors. This data is then uploaded to the host computer core layer, which invokes its built-in media feature library and runs a pattern matching subroutine. The matching process calculates the degree of match between the currently acquired feature vector and a preset media feature vector in the feature library using a specific formula. This is achieved by automatically identifying the type of medium being tested based on pressure fluctuations and temperature rise characteristics. Based on the identification results, the system retrieves the corresponding basic compensation coefficient from the medium feature library and performs iterative optimization calculations using the least squares method based on the initial collected data. Finally, it generates and stores compensation coefficients suitable for the current precise working conditions.

[0161] Step 3: Idle Run Learning and Normal Baseline Establishment

[0162] The system issues commands to control the diaphragm pump to operate stably under no-load conditions for a preset period of time. During this period, the hardware acquisition layer continuously and synchronously collects multi-dimensional parameters such as inlet and outlet pressure, pump body temperature, medium temperature, and operating noise. The edge computing layer processes and analyzes this batch of normal operation data to generate a parameter baseline representing the health status of the pump type. To quantify the stability of the monitoring and learning process, the system calculates the instantaneous deviation of any parameter at any given moment. Meanwhile, the normal data collected during this stage is used to train and optimize the parameters of the lightweight random forest model deployed on the edge computing layer, laying the foundation for subsequent intelligent diagnosis.

[0163] Step 4: Multi-parameter collaborative acquisition and edge preprocessing under test conditions:

[0164] After entering the formal test, the system issues a load test command to put the diaphragm pump into test mode. The hardware acquisition layer simultaneously collects multi-dimensional parameters, including inlet and outlet pressure, pump body temperature, medium temperature, noise, and ambient atmospheric pressure. The collected raw signals are first filtered, amplified, and photoelectrically isolated by the adaptive signal conditioning unit inside the hardware acquisition layer to prevent interference. The processed data is sent to the edge computing layer, which performs noise reduction on the data and extracts the time domain features, including peak value and variance, and the frequency domain features, including main frequency and harmonic amplitude, in real time to generate feature data for intelligent analysis. The edge computing layer also compares the collected data with the preset threshold and the normal parameter baseline generated in step three in milliseconds in real time to quickly filter out abnormal data.

[0165] Step 5: Intelligent Fault Prediction and Digital Twin Visualization:

[0166] The edge computing layer uploads the generated feature data to the host computer core layer. The host computer core layer then inputs this feature data into a pre-trained lightweight random forest model for fault prediction. The model outputs the predicted fault type and its corresponding confidence level. To improve reliability, the system also performs a secondary verification of the model's prediction results using a traditional rule-based threshold method. When the fault confidence level reaches a preset alarm threshold, the host computer core layer triggers a tiered alarm, including a software pop-up and an alarm sound. Simultaneously, the host computer core layer runs a digital twin visualization module, driving a virtual twin model that corresponds one-to-one with the physical pump and includes core virtual components. The parameters collected in real time and the fault prediction results are associated and mapped to this virtual model. When an anomaly occurs, the corresponding virtual component will be highlighted in red and flashing, and its historical data curves can be viewed interactively, achieving visualization and accurate diagnosis of the status.

[0167] Step Six: Remote Security Control Based on Encryption and Redundancy:

[0168] Users with the appropriate operating permissions can send start / stop and emergency stop control commands through the human-machine interface based on the monitoring status. The host computer core layer encrypts the command using an encryption algorithm and transmits it to the edge computing layer through the industrial network. A checksum is added during transmission to ensure the integrity of the command. After the edge computing layer decrypts and verifies the command, it drives the solid-state relays and motor controllers in the execution response layer to perform the corresponding actions. In particular, when an emergency stop command is executed, the hardware emergency stop switch connected in series in the motor's main power supply circuit will act synchronously, forming hardware redundancy protection independent of software logic, forcibly cutting off the power supply, thereby constructing a full-link security control from software encryption to hardware disconnection.

[0169] Step 7: Edge-Cloud Collaborative Data Management and Report Generation

[0170] The system systematically manages all data generated during the testing process, including preset parameters, feature data, and abnormal data. It adopts a collaborative mechanism of edge preprocessing and cloud aggregation storage to classify, store, and query data locally and in the cloud. Finally, the system automatically integrates and analyzes the various data to generate a structured and complete test report, completing the intelligent monitoring and recording of a test cycle.

[0171] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0172] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW, characterized in that, The method includes the following steps: S1. Perform system initialization and two-way authentication of the equipment, and load the preset parameters of the test pump type; S2. Based on preset parameters, perform automatic identification of media type and adaptive calibration of compensation coefficient to generate accurate compensation coefficient; S3. Control the diaphragm pump to run under no-load according to the precise compensation coefficient to collect normal operation parameters, generate the normal parameter baseline of the pump type and complete the training and optimization of the lightweight machine learning model. S4. Based on the machine learning model, perform collaborative acquisition of multi-dimensional parameters and edge preprocessing under test conditions to generate feature data and abnormal data; S5. Based on the feature data and the trained lightweight machine learning model, perform intelligent fault prediction, and combine the digital twin model to realize the visualization mapping and display of the operating status. S6. Based on the operating status and authorized instructions, perform remote security control based on encrypted transmission and hardware redundancy protection; S7. Perform edge and cloud collaborative management on the preset parameters, feature data and abnormal data, and automatically generate a test report.

2. The multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW according to claim 1, characterized in that, In step S1, system initialization and two-way device authentication are performed. Loading the preset parameters for the test pump includes the following steps: S11. After the device is powered on, the edge computing layer and the host computer core layer complete two-way authentication through the pre-stored unique serial number of the device and the dynamic key generated in real time, and refuse unauthorized device access. S12. After certification, the host computer core layer loads the preset parameters corresponding to the current test diaphragm pump model. The preset parameters include rated head, rated flow rate, rated voltage, rated current, rated power, rated input pressure, and rated output pressure.

3. The multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW according to claim 1, characterized in that, The S2 step, which involves automatic identification of the medium type and adaptive calibration of the compensation coefficient based on preset parameters to generate accurate compensation coefficients, includes the following steps: S21. Start the hardware acquisition layer, collect the initial pressure data and initial temperature data of the diaphragm pump at the beginning of startup through pressure and temperature sensors, and transmit them to the edge computing layer. S22. The edge computing layer uploads the initial pressure data and initial temperature data to the host computer core layer. The host computer core layer calls its built-in medium feature library and runs the pattern matching subroutine to automatically identify the current test medium type based on the pressure fluctuation characteristics and temperature rise characteristics. The identification is achieved by calculating the matching degree between the currently acquired feature vector and the preset medium feature vector in the feature library. To achieve this, the calculation formula is: ; in, The value represents the matching degree, ranging from [0, 1]. The closer the value is to 1, the higher the matching degree. The total number of feature dimensions. The current data collection number 3D eigenvalues For a certain preset medium template in the feature library, the first 3D eigenvalues For the first The weight coefficients of the features are pre-set based on the feature's ability to distinguish media types; S23. Based on the identified medium type, retrieve the corresponding basic compensation coefficient from the medium feature library, and based on the initial acquisition data, perform iterative optimization calculations using the least squares method to obtain and store the accurate compensation coefficient applicable to the current test conditions.

4. The multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW according to claim 1, characterized in that, In step S3, controlling the diaphragm pump to run under no-load conditions to collect normal operating parameters, generating a baseline of normal parameters for the pump type, and completing the training and optimization of the lightweight machine learning model includes the following steps: S31. Issue an unloaded operation command to control the diaphragm pump to run stably under unload conditions for a preset time; S32. During no-load operation, multi-dimensional parameters are continuously and synchronously collected through the hardware acquisition layer. The multi-dimensional parameters include inlet and outlet pressure, pump body temperature, medium temperature and operating noise. S33. The edge computing layer processes and analyzes the multi-dimensional parameters collected during no-load operation to generate the parameter baseline of the pump type under normal conditions. The parameter baseline includes the noise baseline, pressure fluctuation baseline and temperature baseline. For any parameter At any point during the baseline self-learning phase instantaneous deviation The calculation formula is: ; in, For parameters At any moment The collected values, For parameters The average value within the collected time period. For parameters The edge computing layer continuously monitors the standard deviation within the collected time period. This ensures that the system operates within a preset range to confirm its stable operation. S34. Simultaneously, the normal operation data collected during the idle operation period is used to train and optimize the parameters of the lightweight random forest model deployed in the edge computing layer.

5. The multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW according to claim 1, characterized in that, The S4 step, which involves collaborative acquisition of multi-dimensional parameters and edge preprocessing based on a machine learning model under test conditions to generate feature data and anomaly data, includes the following steps: S41. Issue a load test command to put the diaphragm pump into test mode; S42. Multi-dimensional parameters under test conditions are synchronously collected through the hardware acquisition layer. The multi-dimensional parameters include inlet and outlet pressure, pump body temperature, medium temperature, noise, and ambient atmospheric pressure. S43. The adaptive signal conditioning unit integrated inside the hardware acquisition layer performs filtering, amplification, and opto-isolation processing on the acquired raw signal. S44. The processed data is transmitted to the edge computing layer, which performs noise reduction on the data and extracts the time-domain and frequency-domain features of the data. The time-domain features include peak value and variance, and the frequency-domain features include the main frequency and harmonic amplitude, generating feature data. S45. The edge computing layer compares the collected data with the preset threshold and the normal parameter baseline in real time, and filters out abnormal data in milliseconds.

6. The multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW according to claim 1, characterized in that, Step S5, which involves intelligent fault prediction based on the feature data and a trained lightweight machine learning model, and the visualization and display of the operational status using a digital twin model, includes the following steps: S51, The edge computing layer uploads the feature data to the host computer core layer; S52. The host computer core layer inputs the received feature data into a trained and optimized lightweight random forest model for fault prediction, and outputs the predicted fault type and corresponding confidence level. S53. The prediction results output by the lightweight random forest model are subjected to secondary verification using a method based on traditional rule thresholds. S54. When the fault confidence level reaches the preset alarm threshold, the host computer core layer triggers a graded alarm, which includes a pop-up notification on the software interface and an alarm sound. S55. At the same time, the host computer core layer runs a digital twin visualization module, which drives a virtual twin model that corresponds one-to-one with the physical diaphragm pump. The virtual twin model includes core virtual components such as the pump body, impeller, diaphragm, and inlet and outlet pipelines. S56. The real-time collected multi-dimensional parameters and the fault prediction results are associated and mapped with the virtual twin model. When the parameters are abnormal and a fault is predicted, the corresponding virtual component is highlighted in red and flashed in the virtual twin model. It also supports viewing the historical parameter curves and fault records of any virtual component through interactive operation.

7. The multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW according to claim 1, characterized in that, The S6 step, based on the operating status and authorized instructions, performs remote security control based on encrypted transmission and hardware redundancy protection, including the following steps: S61. Users with corresponding operating permissions send start / stop control commands and emergency stop control commands downwards through the human-machine interface according to the monitoring status displayed by the host computer core layer. S62. The host computer core layer uses an encryption algorithm to encrypt the control commands and transmits the encrypted commands to the edge computing layer through an industrial communication network. S63. After the edge computing layer decrypts and verifies the command, it drives the solid-state relay and motor controller in the execution response layer to perform the corresponding start and stop actions. S64. When an emergency stop control command is executed, the hardware emergency stop switch connected in series in the motor main power supply circuit will act synchronously to form hardware redundancy protection and forcibly cut off the motor power supply. S65. The execution response layer feeds back the instruction execution results and device status to the edge computing layer in real time, and finally uploads them to the host computer core layer for status update and display.

8. A multi-parameter intelligent monitoring system for a diaphragm pump test platform based on LabVIEW, used to implement the multi-parameter intelligent monitoring method for a diaphragm pump test platform based on LabVIEW as described in any one of claims 1 to 7, characterized in that, The system includes a hardware acquisition layer, an edge computing layer, a host computer core layer, and an execution response layer. The hardware acquisition layer acquires multi-dimensional physical parameters during the diaphragm pump test process and conditions the acquired raw signals. The edge computing layer is communicatively connected to the hardware acquisition layer, and performs local preprocessing, feature extraction, anomaly screening, model inference, data caching, and secure communication on the conditioned data. The host computer core layer is built on the LabVIEW development platform and is remotely connected to the edge computing layer through an industrial communication network. It features media identification and compensation calibration, intelligent fault prediction, full-link visual monitoring, data collaborative management, and security policy control. The execution response layer is connected to the edge computing layer, receives control commands and drives the actuator to perform actions, realizes the start-up and emergency stop control of the diaphragm pump, and has a hardware redundancy protection mechanism.

9. The multi-parameter intelligent monitoring system for a diaphragm pump test platform based on LabVIEW according to claim 8, characterized in that, The host computer core layer includes a media adaptive head calculation module, a lightweight machine learning fault prediction module, a digital twin visualization module, an edge-cloud collaborative data management module, and a full-link security control module; The adaptive head calculation module has a built-in medium feature library. It automatically identifies the test medium type based on pressure and temperature characteristics, and determines the accurate compensation coefficient through iterative optimization algorithm. It calculates high-precision head by combining the collected pressure difference, medium density, gravitational acceleration, position height difference and compensation coefficient. The lightweight machine learning fault prediction module integrates a lightweight random forest model, which predicts fault type and confidence level based on feature data uploaded from the edge computing layer, and performs secondary verification by combining traditional rule thresholds to trigger graded alarms. The digital twin visualization module constructs and drives a virtual twin model that is one-to-one with the physical pump, enabling real-time, intuitive, and relevant visualization and interaction of operating parameters, equipment status, and fault information. The edge-cloud collaborative data management module adopts a collaborative mechanism of edge preprocessing and cloud aggregation storage to manage the local and cloud storage, classification, and query of raw data, feature data, and abnormal data, and automatically generate structured test reports. The full-link security control module manages the security of the entire process from device access authentication, data transmission, user operation permissions to operation log auditing.

10. The multi-parameter intelligent monitoring system for a diaphragm pump test platform based on LabVIEW according to claim 9, characterized in that, The hardware acquisition layer includes a multi-dimensional sensor group and an adaptive signal conditioning unit; the edge computing layer includes a microprocessor core, a local storage unit, a hardware encryption unit, and a feature processing unit; and the execution response layer includes a solid-state relay, a hardware emergency stop switch, and a motor controller with status feedback. The multi-dimensional sensor group includes a pressure sensor for measuring inlet and outlet pressure, a temperature sensor for measuring pump body and medium temperature, a noise sensor for measuring operating noise, and an atmospheric pressure sensor for measuring ambient atmospheric pressure. The adaptive signal conditioning unit integrates a filter circuit, an instrumentation amplifier, and an opto-isolator to perform anti-interference conditioning on the raw signal output by the sensor. The microprocessor core is the main control chip of the edge computing layer, responsible for scheduling and executing data preprocessing, feature extraction, model running and communication control tasks. The local storage unit caches the original data at a high frequency and persists the data when the network is interrupted. The hardware encryption unit performs encryption and decryption operations on uplink data and downlink commands. The feature processing unit extracts time-domain and frequency-domain features from the preprocessed data in real time and constructs feature vectors. The solid-state relay and the motor controller receive instructions from the edge computing layer to control the start and stop of the diaphragm pump motor; The hardware emergency stop switch is physically connected in series in the main power supply circuit of the diaphragm pump motor, serving as a hardware redundant safety protection device independent of software logic.

Images