Control valve system state monitoring method based on LabVIEW and DP-SKNN algorithms

Abstract

The invention relates to the technical field of industrial equipment operation state monitoring, solves the technical problem that a traditional method has limitations in the aspects of real-time performance, monitoring precision and multi-parameter processing capacity, and particularly relates to a control valve system state monitoring method based on LabVIEW and DP-SKNN algorithms. The method comprises the following steps: acquiring key operation parameter data of a control valve by utilizing various sensors, and realizing smooth classification of boundary samples by adopting a soft KNN classification algorithm and introducing Gaussian weight and probability distribution; meanwhile, a key parameter K value of the KNN classifier is optimized by using a particle swarm optimization (PSO) algorithm for dynamically adjusting an inertia weight. And finally, the optimized DP-SKNN algorithm is embedded into a LabVIEW platform. According to the method, the defects of a traditional method in the aspects of real-time performance, classification precision and complex state processing capacity are effectively overcome, the intelligence and reliability of state monitoring of the control valve system are remarkably improved, and the method is suitable for the fields of industrial automation, intelligent manufacturing and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial equipment operation status monitoring, and in particular to a control valve system status monitoring method based on LabVIEW and DP-SKNN algorithm. Background Art

[0002] In recent years, the operation status monitoring of industrial equipment has become one of the important technologies to ensure the stability of the production process and optimize the operation efficiency. As the core actuator of the industrial process control system, the operation status of the control valve directly affects the adjustment accuracy of the process parameters and the overall performance of the system. However, traditional status monitoring methods usually rely on manual inspections or single parameter monitoring equipment, which have problems such as poor real-time performance, limited data processing capabilities, and insufficient intelligence, making it difficult to meet the needs of modern industry for efficient status monitoring under complex working conditions.

[0003] In the field of industrial status monitoring, LabVIEW, as a virtual instrument development platform, is widely used in the status monitoring system of industrial equipment due to its flexible multi-source data acquisition, processing and visualization capabilities. The LabVIEW platform can realize real-time acquisition and analysis of multi-parameters of the operating status of control valves, providing efficient data support for the application of complex algorithms. However, relying solely on the acquisition and processing functions of LabVIEW cannot effectively solve the problem of deep mining and accurate identification of multi-parameter features under complex industrial conditions, especially the need for high-precision monitoring of status modes. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a control valve system status monitoring method based on LabVIEW and DP-SKNN algorithm, which solves the technical problems of limitations of traditional methods in real-time, monitoring accuracy and multi-parameter processing capabilities.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a control valve system state monitoring method based on LabVIEW and DP-SKNN algorithm, the method comprising the following steps:

[0006] S1. Build a LabVIEW data acquisition platform and obtain multi-source operating parameter data of the control valve based on multiple types of sensors;

[0007] S2, pre-processing the multi-source operating parameter data and performing state labeling to obtain training set samples and test set samples consisting of state tags that express the real operating state of the control valve;

[0008] S3, based on the training set samples and the test set samples, a KNN classification model is established to achieve smooth classification of boundary samples in the state classification process;

[0009] S4, using the particle swarm optimization algorithm with dynamic inertia weight adjustment to optimize the classification parameter K of the KNN classification model to obtain the DP-SKNN algorithm model, and the inertia weight is adaptively adjusted according to the change rate of the particle fitness value;

[0010] S5. Embed the DP-SKNN algorithm model into the LabVIEW measurement and control system to realize real-time monitoring of the operating status of the control valve.

[0011] Furthermore, in step S1, the multi-source operating parameter data includes flow signals before and after the control valve, pressure signals before and after the control valve, valve opening signals, and liquid level signals. The multi-source operating parameter data are collected through a variety of sensors and uploaded to the LabVIEW measurement and control system.

[0012] Further, in step S2, the state tag Tag includes a normal state "0", a warning state "1", and a fault state "2", and the state tag Tag is marked according to the control valve operation threshold and historical data.

[0013] Furthermore, in step S3, the specific process includes the following steps:

[0014] S31. Calculate the training sample data x in the training set samples t And the test sample data x in the test set sample i The distance between t , and sort the training set samples by distance, and select the K samples with the closest distance as the test sample data x i Candidate classification reference, that is, candidate sample t, distance d t The calculation formula is:

[0015] d t = Dist(x i ,x t )

[0016] S32, calculate the Gaussian weight γ of candidate sample t t , the calculation formula is:

[0017]

[0018] Among them, γ t is the weighted value of candidate sample t; σ is the smoothing parameter; d i For the test sample data x i The Euclidean distance from the candidate sample t;

[0019] S33, according to the Gaussian weight γ of the candidate sample t t A weighted score is calculated for each status category, and the weighted score formula is:

[0020]

[0021] Among them, N i Tags i Neighbor set of label category; γ t is the Gaussian weight belonging to this category;

[0022] S34. Normalize the weighted score v(Tagi) of each category into a probability distribution P(Tagi|xi). The probability distribution formula is:

[0023]

[0024] S35, according to the probability distribution P(Tag i x i ) determines the final classification result of the test set samples.

[0025] Furthermore, in step S4, the specific process includes the following steps:

[0026] S41, initialize the particle swarm, randomly generate a group of particles, each particle represents a K value, K i represents the K value corresponding to the i-th particle;

[0027] S42, establish a fitness function, divide the training set samples into 5 subsets of similar size, cyclically use different subsets as validation sets, and the remaining subsets as training sets, so that each training data sample is used as a validation sample at least once to calculate the classification accuracy and as the fitness value. The expression of the fitness function is:

[0028]

[0029] Among them, Fitness (K i ) represents the fitness value calculated by the i-th particle; n d is the number of training data samples; Accuracy(K i ) represents the accuracy of the K value calculation corresponding to the i-th particle;

[0030] S43, adaptively adjusting the inertia weight of the particle swarm optimization algorithm according to the change rate of the particle fitness value, the expression for dynamically adjusting the inertia weight is:

[0031]

[0032] Where w(t) is the current iteration number, and the value of the first iteration is 1; T max is the maximum number of iterations; w max is the initial inertia weight; w min is the termination inertia weight;

[0033] S44, compare the current fitness value of each particle with the historical optimal fitness value, update the individual optimal position pBest and the global optimal position gBest, and calculate the new speed and position of each particle to obtain a new K value. The speed update and position update expressions are:

[0034] v i (t+1)=w(t)·v i (t)+c 1 ·r 1 ·(pBest i -K i (t))+c 2 ·r 2 ·(gBest i -K i (t))

[0035] K i (t+1)=K i (t)+v i (t+1)

[0036] Among them, v i (t) is the velocity of the ith particle at the tth iteration; w is the inertia weight, which controls the search range of the particle in the search space; c 1 、c 2 The learning factor controls the degree to which the particle moves to the individual optimal position and the global optimal position; r 1 、r 2 is a random number; pBest i With gBest i , are the optimal position of the i-th particle and the global optimal position of the entire particle swarm respectively; K i (t) is the position of the i-th particle at the t-th iteration;

[0037] S45. Repeat step S44 until convergence and output gBest, that is, the optimal classification parameter K.

[0038] The technical solution also provides a system applied to the above-mentioned control valve system status monitoring method, including a control valve system device on which multiple types of sensors are arranged, a LabVIEW measurement and control system for real-time acquisition of multi-source operating parameter data of multiple types of sensors and for data transmission and interaction, and a server end that obtains control valve system device status information based on the multi-source operating parameter data and transmits the status information to the front-end status display interface of the LabVIEW measurement and control system for display.

[0039] Furthermore, the specific process of the LabVIEW measurement and control system to achieve data transmission and interaction is as follows:

[0040] The control valve system device collects multi-source data through the sensor component and uploads it to the database in the LabVIEW measurement and control system through the data acquisition platform;

[0041] The LabVIEW measurement and control system uploads the collected data to the database interaction layer of the server through the TCP / IP protocol;

[0042] The server classifies and analyzes the received data through the data processing module, and feeds back the classification results to the data interaction layer of the LabVIEW measurement and control system through the TCP / IP protocol;

[0043] The LabVIEW measurement and control system displays the equipment operating status in real time through the front-end status display interface, and sends feedback signals to the control valve device through the data interaction layer to adjust the operating parameters of the control valve system device.

[0044] By means of the above technical solution, the present invention provides a control valve system state monitoring method based on LabVIEW and DP-SKNN algorithm, which has at least the following beneficial effects:

[0045] 1. The present invention can realize the real-time acquisition of multi-parameter data of the operating status of the control valve and the efficient application of the DP-SKNN algorithm, combining the advantages of the intelligent algorithm with the flexibility of the virtual instrument, and can provide accurate status monitoring results under complex working conditions. It provides an efficient and reliable solution for the intelligent status monitoring of modern industrial equipment, and has important research value and broad application prospects.

[0046] 2. The present invention effectively solves the deficiencies of traditional methods in terms of real-time performance, classification accuracy and complex state processing capabilities, significantly improves the intelligence and reliability of control valve system status monitoring, and is suitable for fields such as industrial automation and intelligent manufacturing.

[0047] 3. The present invention introduces a soft classification strategy to assign category probability distribution to samples, thereby enhancing the adaptability to boundary samples and complex state patterns.

[0048] 4. The present invention optimizes the key parameter K by dynamically adjusting the PSO algorithm of the inertia weight, thereby solving the deficiency of the traditional algorithm in the balance between global search and local search, and effectively improving the efficiency and accuracy of parameter optimization.

[0049] 5. The particle swarm optimization algorithm PSO proposed in the present invention is combined with soft KNN to form the DP-SKNN algorithm, which realizes the synergy of parameter optimization and state classification, and can more efficiently process multi-source monitoring data and accurately identify the operating status of the control valve. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0051] Figure 1 It is a flow chart of the control valve system status monitoring method in the present invention;

[0052] Figure 2 This is a schematic diagram of the traditional KNN voting mechanism in the present invention;

[0053] Figure 3 Improved KNN algorithm flow chart for the present invention;

[0054] Figure 4 It is a schematic diagram for comparing the state monitoring results in the present invention;

[0055] Figure 5 It is a schematic diagram of the control valve system status monitoring system in the present invention. DETAILED DESCRIPTION

[0056] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific implementation methods, so that the implementation process of how the present application uses technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly.

[0057] In order to effectively solve the problem of deep mining and accurate identification of multi-parameter features under complex industrial conditions, especially the need for high-precision monitoring of state modes, please refer to Figure 1-Figure 4 This embodiment proposes a control valve system status monitoring method based on LabVIEW and DP-SKNN algorithm, which can realize the real-time acquisition of multi-parameter data of the control valve operating status and the efficient application of DP-SKNN algorithm, combining the advantages of intelligent algorithms with the flexibility of virtual instruments, and can provide accurate status monitoring results under complex working conditions, providing an efficient and reliable solution for the intelligent status monitoring of modern industrial equipment, which has important research value and broad application prospects. Figure 1 As shown, the method comprises the following steps:

[0058] S1. Build a LabVIEW data acquisition platform, and obtain multi-source operating parameter data of the control valve based on multiple types of sensors, wherein the multi-source operating parameter data includes flow signals before and after the control valve, pressure signals before and after the control valve, valve opening signals, and liquid level signals. The multi-source operating parameter data is collected through multiple sensors and uploaded to the LabVIEW measurement and control system. The control valve experimental equipment used in this embodiment is mainly equipped with the following sensor types: turbine flowmeter, diffused silicon pressure transmitter, diffused silicon liquid level transmitter, valve position opening sensor, a total of 4 sensors, of which the turbine flowmeter is divided into a pre-valve flow sensor and a post-valve flow sensor, and the diffused silicon pressure sensor is divided into a pre-valve pressure sensor and a post-valve pressure sensor, a total of 6 sensor elements.

[0059] The data acquisition board is used to perform multi-channel data acquisition tasks for various sensors designed in the control valve system. The main collected signals include: flow signals before and after the control valve, pressure signals before and after the control valve, valve opening signals, and liquid level signals. For the signal mode of the main collected signals, the current analog signal acquisition board and the voltage analog signal acquisition board are selected.

[0060] According to the output signal type and signal range of the sensor, the PCle-4303-C acquisition board produced by NI was selected, and the acquisition card AI0-5 acquisition channels were connected respectively. By calling the package function NIMAX supported by the LabVIEW platform, the host computer program was written to transmit the real-time acquired signal to the front-end display, and the data was stored in the TDMS file format unique to the LabVIEW platform and dedicated to measurement and control data storage, forming a data set for training and testing.

[0061] S2. After preprocessing the multi-source operating parameter data, the state is annotated to obtain training set samples and test set samples consisting of state tags that express the real operating state of the control valve. The state tags include normal state "0", warning state "1", and fault state "2". The state tags are annotated according to the control valve operating threshold and historical data. The preprocessing of the multi-source operating parameter data includes data preprocessing methods such as filtering and noise reduction. The new sensor data x generated t =[x 1 (t),x 2 (t),x 3 (t),...,x n (t)], where x i (t) represents the output data of the i-th sensor at time t, and n represents the number of sensors.

[0062] In this embodiment, according to the actual operating conditions of the control valve system, the multi-source operating parameter data collected by the sensor is marked with a state tag that can express the actual operating state of the control valve system, including sensor data with known states and training set data, which can be expressed as:

[0063] x t =[x 1 (t),x 2 (t),x 3 (t),...,x n (t),y t ]

[0064] X=[x 1 x 2 ,...,x n ] T

[0065] where y t Represents the data label at time t.

[0066] Table 1 Description of sensor variables in control valve system

[0067]

[0068] According to Table 1, the sensor data set of the control valve system device can be specifically described as:

[0069]

[0070] Among them, x i Represents the i-th sample row vector containing 6 sensor data.

[0071] Through the simulation state experiment, the characteristic data of the sensor under different states are recorded, and the equipment status is divided into normal state, warning state, and fault state. The division criteria are shown in Table 2.

[0072] Table 2 Description of status labels of control valve system

[0073]

[0074]

[0075] According to the above-mentioned criteria, the collected training set samples X are marked with corresponding label data, namely:

[0076] Y=[y 1 y 2 ,....,y n ] T

[0077] Among them, Y is the column vector corresponding to the training set sample X, y i is the label value of the corresponding sample.

[0078] like Figure 2 As shown in the figure, the parameter K of the number of nearest neighbor samples used by the traditional KNN classifier to determine the classification determines how many nearest neighbors are selected in the data set to help make classification decisions. Therefore, the traditional KNN is a hard decision rule, and its decision-making method is to sort the training samples from near to far, and obtain the classification result through a voting mechanism.

[0079] S3. A KNN classification model is established based on the training set samples and the test set samples to realize smooth classification of boundary samples in the state classification process. In order to cope with the irregular decision boundary of the traditional KNN, which may lead to overfitting when there is a lot of data noise or a small K value, this embodiment proposes a KNN classification model of a soft KNN classifier, such as Figure 3 As shown, the main implementation steps are as follows:

[0080] S31. Calculate the training sample data x in the training set samples t And the test sample data x in the test set sample i The distance between t , and sort the training set samples by distance, and select the K samples with the closest distance as the test sample data x i Candidate classification reference, that is, candidate sample t, distance d t The calculation formula is:

[0081] d t = Dist(x i ,x t )

[0082] In this embodiment, the KNN nearest neighbor algorithm classification process needs to calculate the Euclidean distance between the test sample data and each data in the training set sample, and the calculation formula is:

[0083]

[0084] Among them, x j (t) is the data to be classified x t The jth sensor value of j (i) is the data to be classified x i The jth sensor value of t As training sample data, x i As test sample data.

[0085] S32, calculate the Gaussian weight γ of candidate sample t t , the calculation formula is:

[0086]

[0087] Among them, γ t is the weighted value of candidate sample t; σ is the smoothing parameter; d i For the test sample data x i The Euclidean distance from the candidate sample t;

[0088] S33, according to the Gaussian weight γ of the candidate sample t t A weighted score is calculated for each status category, and the weighted score formula is:

[0089]

[0090] Among them, N i Tags i Neighbor set of label category; γ t is the Gaussian weight belonging to this category;

[0091] S34, to generate a smooth probability distribution, the weighted score v(Tag i ) is standardized to the probability distribution P(Tag i x i ), the probability distribution formula is:

[0092]

[0093] S35, according to the probability distribution P(Tag i x i ) determines the final classification result of the test set samples. Specifically, in the classification of normal state, warning state, and fault state, the probability distribution P(Tag i x i ) is the maximum value, the classification result is normal, and the same applies to other states.

[0094] This embodiment introduces a soft classification strategy to assign category probability distribution to samples, thereby enhancing the adaptability to boundary samples and complex state patterns.

[0095] S4, using the particle swarm optimization algorithm with dynamic inertia weight adjustment to optimize the classification parameter K of the KNN classification model to obtain the DP-SKNN algorithm model, and the inertia weight is adaptively adjusted according to the rate of change of the particle fitness value; in order to improve the accuracy of the KNN classification model for the classification results, this embodiment uses the improved PSO particle swarm optimization algorithm to optimize the key parameter K of the KNN classification model. In step S4, the specific process includes the following steps:

[0096] S41, initialize the particle swarm, randomly generate a group of particles, each particle represents a K value, K i represents the K value corresponding to the i-th particle;

[0097] S42, establish a fitness function, divide the training set samples into 5 subsets of similar size, cyclically use different subsets as validation sets, and the remaining subsets as training sets, so that each training data sample is used as a validation sample at least once to calculate the classification accuracy and as the fitness value. The expression of the fitness function is:

[0098]

[0099] Among them, Fitness (K i ) represents the fitness value calculated by the i-th particle; n d is the number of training data samples; Accuracy(K i ) represents the accuracy of the K value calculation corresponding to the i-th particle;

[0100] S43, according to the change rate of the particle fitness value, the inertia weight of the particle swarm optimization algorithm is adaptively adjusted. In order to improve the rapidity of its optimization convergence and avoid the local optimal situation, its weight number w(t) is dynamically adjusted before each iteration to enhance its local search ability. Therefore, the expression for dynamically adjusting the inertia weight is:

[0101]

[0102] Where w(t) is the current iteration number, and the value of the first iteration is 1; T max is the maximum number of iterations; w max is the initial inertia weight, taking a larger weight of 0.9; w min To terminate the inertia weight, take a smaller weight of 0.1;

[0103] S44, compare the current fitness value of each particle with the historical optimal fitness value, update the individual optimal position pBest and the global optimal position gBest, calculate the new speed and position of each particle according to the speed update formula and the position update formula to obtain a new K value, and the speed update and position update expressions are:

[0104] v i (t+1)=w(t)·v i (t)+c 1 ·r 1 ·(pBest i -K i (t))+c 2 ·r 2 ·(gBest i -K i (t))

[0105] K i (t+1)=K i (t)+v i (t+1)

[0106] Among them, v i (t) is the velocity of the ith particle at the tth iteration; w is the inertia weight, which controls the search range of the particle in the search space; c 1 、c 2 The learning factor controls the degree to which the particle moves to the individual optimal position and the global optimal position; r 1 、r 2 is a random number; pBest i With gBest i , are the optimal position of the i-th particle and the global optimal position of the entire particle swarm respectively; K i (t) is the position of the i-th particle at the t-th iteration;

[0107] S45. Repeat step S44 until convergence and output gBest, that is, the optimal classification parameter K.

[0108] This embodiment optimizes the key parameter K by dynamically adjusting the PSO algorithm of the inertia weight, thereby solving the deficiency of the traditional algorithm in the balance between global search and local search, and effectively improving the efficiency and accuracy of parameter optimization.

[0109] S5. Embed the DP-SKNN algorithm model into the LabVIEW measurement and control system to realize real-time monitoring of the operating status of the control valve. In this embodiment, the optimal classification parameter K and its corresponding accuracy are obtained by the optimization method of step S4, and the optimized DP-SKNN algorithm model is reconstructed, such as Figure 5 As shown, the data acquisition device is connected to the desktop host through the LAN interface, and the two hosts are divided into the client and the server through the TCP / IP protocol. The data acquisition device acts as a client to collect data and then transmits it to the server through the port, that is, the host that builds the DP-SKNN algorithm model optimized by the improved PSO algorithm.

[0110] Then, the collected multi-source operating parameter data is classified by the KNN model to obtain the output label sample type. The client and server hosts exchange roles, and the host equipped with the DP-SKNN algorithm model acts as a client to transmit the label type data to the data acquisition device as the server, and displays the device status results through the LabVIEW host program.

[0111] This embodiment combines virtual instrument technology with intelligent algorithms to achieve real-time monitoring and accurate determination of the operating status of the control valve system under complex working conditions. The system first builds a data acquisition platform based on LabVIEW, and uses a variety of sensors to collect key operating parameter data of the control valve, including flow, pressure, valve opening, and liquid level signals before and after the valve. After the collected data is pre-processed by filtering and noise reduction, some of the data is labeled to reflect the operating status of the control valve, which is used to construct training data sets and test data sets.

[0112] In the state classification process, the soft KNN classification algorithm is used to achieve smooth classification of boundary samples by introducing Gaussian weights and probability distribution; at the same time, the particle swarm optimization algorithm (PSO) with dynamic adjustment of inertia weight is used to optimize the key parameter K value of the KNN classifier to improve the accuracy and robustness of the classifier. Finally, the optimized DP-SKNN algorithm is embedded in the LabVIEW platform, and the data transmission between the host and the acquisition device is realized through the TCP / IP communication protocol, and the operating status of the control valve system is displayed in real time through the LabVIEW interface. The present invention effectively solves the shortcomings of traditional methods in real-time, classification accuracy and complex state processing capabilities, significantly improves the intelligence and reliability of control valve system state monitoring, and is suitable for industrial automation and intelligent manufacturing and other fields.

[0113] This embodiment obtains a sample set of 10082 known state labels through experiments, and obtains the best accuracy of 0.97 by improving the PSO algorithm optimization, corresponding to the key parameter K of the DP-SKNN algorithm model is 328. 89 verification data samples are input, and the state monitoring results are compared. Figure 4 As shown, the accuracy of the recognition result is 0.996. The results of this embodiment verify the effectiveness of the method.

[0114] Embodiment 2

[0115] Based on the first embodiment, this embodiment further proposes a system corresponding to the control valve system state monitoring method in the first embodiment, such as Figure 5 As shown, it includes a control valve system device on which multiple types of sensors are arranged, a LabVIEW measurement and control system for real-time acquisition of multi-source operating parameter data of multiple types of sensors and data transmission and interaction, and a server end that obtains control valve system device status information based on the multi-source operating parameter data and transmits the status information to the front-end status display interface of the LabVIEW measurement and control system for display.

[0116] Specifically, Figure 5As shown in the figure, the LabVIEW measurement and control system transmits sensor data to the data acquisition platform by writing a data acquisition program, and stores the data in the database in the form of a TDMS file format. Through Ethernet connection, a data interaction layer based on the TCP / IP protocol is built, and the data file is transmitted to the data processing module on the server side for training, testing, and judgment, and the status information of the control valve system device is obtained. The status information is transmitted to the front-end status display interface of the LabVIEW measurement and control system through the data feedback of the server side for display.

[0117] Specifically, the LabVIEW measurement and control system realizes data transmission and interaction in the following ways. The specific process is as follows:

[0118] The control valve system device collects multi-source data through the sensor component and uploads it to the database in the LabVIEW measurement and control system through the data acquisition platform;

[0119] The LabVIEW measurement and control system uploads the collected data to the database interaction layer of the server through the TCP / IP protocol;

[0120] The server classifies and analyzes the received data through the data processing module, and feeds back the classification results to the data interaction layer of the LabVIEW measurement and control system through the TCP / IP protocol;

[0121] The LabVIEW measurement and control system displays the equipment operating status in real time through the front-end status display interface, and sends feedback signals to the control valve device through the data interaction layer to adjust the operating parameters of the control valve system device.

[0122] In this embodiment, according to the output signal type and signal range of the sensor, the PCle-4303-C acquisition board produced by NI is selected, and the acquisition card AI0-5 acquisition channels are connected respectively. By calling the package function NIMAX supported by the LabVIEW platform, the host computer program is written to transmit the real-time acquired signal to the front-end display, and the data is stored in the TDMS file format unique to the LabVIEW platform and dedicated to measurement and control data storage, forming a data set for training and testing.

[0123] The present invention proposes a control valve state monitoring method based on a particle swarm optimization algorithm (PSO) with dynamic adjustment of inertia weight and a combination of soft KNN (DP-SKNN). By optimizing the key parameter K through the PSO algorithm with dynamic adjustment of inertia weight, the deficiencies of the traditional algorithm in the balance between global search and local search are solved, and the efficiency and accuracy of parameter optimization are effectively improved. The soft KNN algorithm assigns category probability distribution to samples by introducing a soft classification strategy, thereby enhancing the adaptability to boundary samples and complex state patterns. The combination of the two algorithms realizes the synergy of parameter optimization and state classification, and can more efficiently process multi-source monitoring data and accurately identify the operating status of the control valve.

[0124] Through the integration of the LabVIEW platform, the present invention realizes the real-time acquisition of multi-parameter data of the operating status of the control valve and the efficient application of the DP-SKNN algorithm, combining the advantages of the intelligent algorithm with the flexibility of the virtual instrument, and can provide accurate status monitoring results under complex working conditions. The present invention solves the limitations of traditional methods in terms of real-time performance, monitoring accuracy and multi-parameter processing capabilities, and provides an efficient and reliable solution for the intelligent status monitoring of modern industrial equipment, which has important research value and broad application prospects.

[0125] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, so the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0126] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the above embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0127] The above implementation methods have been described in detail. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A control valve system status monitoring method based on LabVIEW and DP-SKNN algorithm, characterized in that: The method comprises the following steps: S1. Build a LabVIEW data acquisition platform and obtain multi-source operating parameter data of the control valve based on multiple types of sensors; S2, pre-processing the multi-source operating parameter data and performing state labeling to obtain training set samples and test set samples consisting of state tags that express the real operating state of the control valve; S3, based on the training set samples and the test set samples, a KNN classification model is established to achieve smooth classification of boundary samples in the state classification process; S4, using the particle swarm optimization algorithm with dynamic inertia weight adjustment to optimize the classification parameter K of the KNN classification model to obtain the DP-SKNN algorithm model, and the inertia weight is adaptively adjusted according to the change rate of the particle fitness value; S5. Embed the DP-SKNN algorithm model into the LabVIEW measurement and control system to realize real-time monitoring of the operating status of the control valve.

2. The control valve system status monitoring method according to claim 1, characterized in that: In step S1, the multi-source operating parameter data includes flow signals before and after the control valve, pressure signals before and after the control valve, valve opening signals, and liquid level signals. The multi-source operating parameter data are collected through various sensors and uploaded to the LabVIEW measurement and control system.

3. The control valve system status monitoring method according to claim 1, characterized in that: In step S2, the state tag Tag includes a normal state "0", a warning state "1", and a fault state "2", and the state tag Tag is marked according to the control valve operation threshold and historical data.

4. The control valve system status monitoring method according to claim 1, characterized in that: In step S3, the specific process includes the following steps: S31. Calculate the training sample data x in the training set samples t The distance d between the test sample data and the test set sample t , and sort the training set samples by distance, and select the K samples with the closest distance as the test sample data x i Candidate classification reference, that is, candidate sample t, distance d t The calculation formula is: d t =Dist(x i ,x t ) S32. Calculate the Gaussian weight γ of candidate sample t t , the calculation formula is: Among them, γ t is the weighted value of candidate sample t; σ is the smoothing parameter; d i For the test sample data x i The Euclidean distance from the candidate sample t; S33, according to the Gaussian weight γ of the candidate sample t t A weighted score is calculated for each status category, and the weighted score formula is: Among them, N i Tags i Neighbor set of label category; γ t is the Gaussian weight belonging to this category; S34, weighted score v(Tag i ) is standardized to the probability distribution P(Tag i x i ), the probability distribution formula is: S35. Determine the final classification result of the test set samples according to the maximum value of the probability distribution P(Tagi|xi).

5. The control valve system status monitoring method according to claim 1, characterized in that: In step S4, the specific process includes the following steps: S41, initialize the particle swarm, randomly generate a group of particles, each particle represents a K value, K i represents the K value corresponding to the i-th particle; S42, establish a fitness function, divide the training set samples into 5 subsets of similar size, cyclically use different subsets as validation sets, and the remaining subsets as training sets, so that each training data sample is used as a validation sample at least once to calculate the classification accuracy and as the fitness value. The expression of the fitness function is: Among them, Fitness(K i ) represents the fitness value calculated by the i-th particle; n d is the number of training data samples; Accuracy(K i ) represents the accuracy of the K value calculation corresponding to the i-th particle; S43, adaptively adjusting the inertia weight of the particle swarm optimization algorithm according to the change rate of the particle fitness value, the expression for dynamically adjusting the inertia weight is: Where w(t) is the current iteration number, and the value of the first iteration is 1; T max is the maximum number of iterations; w max is the initial inertia weight; w min is the termination inertia weight; S44, compare the current fitness value of each particle with the historical optimal fitness value, update the individual optimal position pBest and the global optimal position gBest, and calculate the new speed and position of each particle to obtain a new K value. The speed update and position update expressions are: v i (t+1)=w(t)·v i (t)+c1·r1·(pBest i -K i (t))+c2·r2·(gBest i -K i (t)) K i (t+1)=K i (t)+v i (t+1) Among them, v i (t) is the velocity of the ith particle at the tth iteration; w is the inertia weight, which controls the search range of the particle in the search space; c1 and c2 are learning factors that control the degree to which the particle moves to the individual optimal position and the global optimal position; r1 and r2 are random numbers; pBest i With gBest i , are the optimal position of the i-th particle and the global optimal position of the entire particle swarm respectively; K i (t) is the position of the i-th particle at the t-th iteration; S45. Repeat step S44 until convergence and output gBest, that is, the optimal classification parameter K.

6. A system applied to the control valve system status monitoring method according to any one of claims 1 to 5, characterized in that: It includes a control valve system device on which multiple types of sensors are arranged, a LabVIEW measurement and control system that collects multi-source operating parameter data of multiple types of sensors in real time and transmits and interacts with the data, and a server that obtains status information of the control valve system device based on the multi-source operating parameter data and transmits the status information to the front-end status display interface of the LabVIEW measurement and control system for display.

7. The system according to claim 6, characterized in that The specific process of the LabVIEW measurement and control system to achieve data transmission and interaction is as follows: The control valve system device collects multi-source data through the sensor component and uploads it to the database in the LabVIEW measurement and control system through the data acquisition platform; The LabVIEW measurement and control system uploads the collected data to the database interaction layer of the server through the TCP / IP protocol; The server classifies and analyzes the received data through the data processing module, and feeds back the classification results to the data interaction layer of the LabVIEW measurement and control system through the TCP / IP protocol; The LabVIEW measurement and control system displays the equipment operating status in real time through the front-end status display interface, and sends feedback signals to the control valve device through the data interaction layer to adjust the operating parameters of the control valve system device.

Images