Slurry pump pipeline flow adaptive monitoring system and method based on machine learning

Abstract

The present invention discloses a slurry pump pipeline flow adaptive monitoring system and method based on machine learning, which relates to the field of flow monitoring. The present invention monitors and maps data in a slurry pump pipeline from various types of data such as concentration, particle size distribution, pressure, temperature, etc., and makes the mapping more accurate and comprehensive by considering various factors; by constructing a final slurry pump pipeline flow mapping equation, the fluid data of multiple slurry pump pipelines to be monitored that are subsequently collected in real time are mapped, thereby obtaining corresponding flow data, and further providing data support for ultimately determining whether the flow data of the slurry pump pipeline to be monitored currently meets the requirements; by predicting the slurry pump pipeline fluid data at future moments and obtaining corresponding flow data, it is possible to determine whether there is a problem with the current motor frequency of the slurry pump to be monitored, and if there is a problem, it can be adjusted in advance to ensure that the slurry pump can work normally and stably.

Description

Technical Field

[0001] The present invention belongs to the field of flow monitoring, and in particular, relates to a slurry pump pipeline flow adaptive monitoring system and method based on machine learning. Background Art

[0002] The flow data of the pipeline in the slurry pump is relatively important. The flow data of the pipeline is closely related to the pressure, temperature and particle distribution therein. If the flow data of the slurry pump pipeline cannot be monitored in real time, it will be impossible to perceive the abnormality of the flow data in the slurry pump pipeline in time, which may lead to a slowdown in the fluid delivery speed, directly affecting the continuity and efficiency of the production process, and affecting the quality of the product, and even causing equipment damage, causing great losses to the enterprise. Summary of the invention

[0003] In response to the problems in the related technology, the present invention proposes a slurry pump pipeline flow adaptive monitoring system and method based on machine learning to overcome the above-mentioned technical problems existing in the existing related technology.

[0004] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0005] The present invention is a method for adaptively monitoring flow in a slurry pump pipeline based on machine learning, comprising the following steps:

[0006] S1, collecting data in the pipeline of the slurry pump to be monitored at multiple time points to obtain a current monitoring data matrix set, a current monitoring image data matrix and a current flow data matrix;

[0007] S2, collecting multiple historical fluid image data in the slurry pump pipeline and corresponding particle distribution feature data in cooperation with the current monitoring image data matrix and constructing a fluid image feature data extraction model;

[0008] S3, using the fluid image feature data extraction model to extract particle distribution features from each monitoring image data in the current monitoring image data matrix, and constructing the final slurry pump pipeline flow mapping equation in conjunction with the current monitoring data matrix set and the current flow data matrix;

[0009] S4, collecting data in the pipeline of the slurry pump to be monitored in real time for multiple times and extracting the particle distribution characteristics of the image data using a fluid image feature data extraction model, and predicting the fluid data of the slurry pump pipeline at future times with the extracted particle distribution feature data to obtain a future monitoring data matrix set and a future monitoring image feature data matrix set;

[0010] S5. Use the final slurry pump pipeline flow mapping equation to map and calculate the future monitoring data matrix set and the future monitoring image feature data matrix set to obtain the future pipeline overall flow data set; adjust the frequency of the slurry pump motor according to the future pipeline overall flow data set to adjust the flow in the slurry pump pipeline;

[0011] By collecting data in the pipeline of the slurry pump to be monitored at multiple time points, data support is provided for the subsequent construction of the final slurry pump pipeline flow mapping equation that can map the flow data in the pipeline; by constructing a fluid image feature data extraction model, it is used to extract the particle distribution feature data from the fluid image data in the slurry pump pipeline to be monitored, which provides particle distribution data support for the subsequent construction of the final slurry pump pipeline flow mapping equation; by constructing the final slurry pump pipeline flow mapping equation, the fluid data of multiple slurry pump pipelines to be monitored collected in real time are subsequently mapped, so as to obtain the corresponding flow data, and then provide data support for the final determination of whether the flow data of the slurry pump pipeline to be monitored meets the requirements; by collecting data in the pipeline of the slurry pump to be monitored in real time for multiple times and using the fluid image feature data extraction model to extract the particle distribution characteristics of the image data therein, data support is provided for the subsequent determination of whether the motor frequency of the slurry pump needs to be adjusted; by predicting the slurry pump pipeline fluid data at future times and obtaining the corresponding flow data, it can be determined whether there is a problem with the current motor frequency of the slurry pump to be monitored. If there is a problem, it can be adjusted in advance to ensure that the slurry pump can work normally and stably.

[0012] Preferably, the S1 comprises the following steps:

[0013] S11, setting a slurry pump pipeline to be monitored; evenly setting a number of monitoring nodes in the slurry pump pipeline to be monitored to obtain a monitoring node set a 1i represents the i-th monitoring node set in the pipeline of the slurry pump to be monitored, and a′1 represents the total number of monitoring nodes set in the pipeline of the slurry pump to be monitored; set a variety of sensor types for collecting data for monitoring flow and sensor types for collecting images, and obtain a data monitoring sensor type set and image monitoring sensor type, a 2i represents the setting of the i-th data monitoring sensor type, a′2 represents the total number of the set data monitoring sensor types;

[0014] S12, evenly set multiple data collection time points to obtain a real-time data collection time point set; arrange sensors at each monitoring node in the monitoring node set according to the data monitoring sensor type set, the image monitoring sensor type and the real-time data collection time point set, and collect data in the pipeline of the slurry pump to be monitored after the slurry pump starts working, to obtain a current monitoring data matrix set Current monitoring image data matrix And the current traffic data matrix represents the data matrix monitored at the i-th time point in the data real-time acquisition time point set, Indicates the total number of set data collection time points; They are as follows:

[0015]

[0016] in, express The data collected by the k-th type of data monitoring sensor at the j-th monitoring node, represents the image data collected at the jth monitoring node at the i-th data collection time point, represents the flow data collected at the jth monitoring node at the i-th data collection time point;

[0017] Data monitoring sensor types include MEMS flow rate sensors, pressure sensors, position sensors, temperature sensors and optical sensors (used to monitor concentration), etc.; the current flow data matrix can be collected using Doppler ultrasonic flowmeters, etc.; image monitoring sensor types include microflow imaging particle analyzers, etc.

[0018] Preferably, S2 comprises the following steps:

[0019] S21, setting multiple particle distribution feature types of the fluid image in the slurry pump pipeline, obtaining a fluid image feature type set b = {b1, ..., b i ,...,b b′}, b i represents the i-th particle distribution feature type of the set fluid image in the slurry pump pipeline, and b′ represents the total number of particle distribution feature types of the set fluid image in the slurry pump pipeline;

[0020] According to the fluid image feature type set, multiple fluid image data in the slurry pump pipeline and corresponding particle distribution feature data are collected to obtain a historical fluid original image data set. And the historical fluid image feature data matrix represents the i-th historical fluid original image data collected, Indicates the total amount of historical fluid raw image data collected; as follows,

[0021]

[0022] in, Represents the j-th type of particle distribution characteristic data of the i-th historical fluid original image data collected;

[0023] S22, constructing a fluid image feature data extraction model using the historical fluid original image data set and the historical fluid image feature data matrix;

[0024] The types of fluid image features include the distribution density of particles in the pipeline, the size range of particles, the shape characteristics of particles, such as spherical, elliptical, irregular shapes, etc., the movement path and speed of particles in the pipeline, the number and frequency of collisions between particles, the deposition position and degree of particles in the pipeline, etc.; by using historical fluid original image data sets and historical fluid image feature data matrices to construct a fluid image feature data extraction model, the obtained fluid image feature data extraction model has a better extraction effect on the feature data of fluid images.

[0025] Preferably, S3 comprises the following steps:

[0026] S31, input each image data in the current monitoring image data matrix into the fluid image feature data extraction model to extract its particle distribution feature data, and obtain the current monitoring image feature data matrix set It represents the particle distribution characteristic data matrix of the fluid in the pipeline of the slurry pump to be monitored at the i-th time point in the real-time data collection time point set; as follows,

[0027]

[0028] in, express The k-th type of particle distribution characteristic data of the image data collected at the j-th monitoring node;

[0029] S32, constructing a final slurry pump pipeline flow mapping equation using the current monitoring data matrix set, the current monitoring image feature data matrix set and the current flow data matrix;

[0030] By inputting each image data in the current monitoring image data matrix into the fluid image feature data extraction model to extract its particle distribution feature data, the particle distribution parameter data support is provided for the subsequent construction of the final slurry pump pipeline flow mapping equation, thereby facilitating the construction of a mapping equation related to the particle distribution parameter data.

[0031] Preferably, the S32 comprises the following steps:

[0032] S321, constructing the initial slurry pump pipeline flow mapping equation; as follows,

[0033]

[0034] Wherein, c is the dependent variable of the initial slurry pump pipeline flow mapping equation, which represents the slurry pump pipeline flow data; c′ 2i , c′ 4i are the ith independent variable and the (a′2+i)th independent variable of the slurry pump pipeline flow mapping equation, respectively, representing the data collected by the ith type of data monitoring sensor and the ith type of particle distribution characteristic data, c1′ i , c3′ i c′ 2i , c′ 4i The independent variable coefficient; c5′ represents the offset of the initial slurry pump pipeline flow mapping equation;

[0035] S322, each monitoring data set in the current monitoring data matrix set and the current monitoring image feature data matrix set and the corresponding monitoring image feature data set are combined and substituted into the initial slurry pump pipeline flow mapping equation for mapping, and the current flow mapping data matrix is ​​obtained. as follows,

[0036]

[0037] in, Indicates that The jth row of data in Substitute the j-th row data in into the initial slurry pump pipeline flow mapping equation to obtain the flow mapping data;

[0038] S323, setting a mapping Euclidean distance threshold; calculating the Euclidean distance between the current traffic mapping data matrix and the current traffic data matrix to obtain a mapping Euclidean distance The calculation formula is as follows,

[0039]

[0040] When the mapped Euclidean distance is less than the mapped Euclidean distance threshold, there is no need to optimize and adjust the initial slurry pump pipeline flow mapping equation; otherwise, the initial slurry pump pipeline flow mapping equation is optimized and adjusted until the mapped Euclidean distance is less than the mapped Euclidean distance threshold;

[0041] The current monitoring data matrix set and each monitoring data set in the current monitoring image feature data matrix set and the corresponding monitoring image feature data set are combined and substituted into the initial slurry pump pipeline flow mapping equation for mapping. Since the independent variable coefficients and bias of the initial slurry pump pipeline flow mapping equation are randomly selected, it is necessary to test the mapping accuracy of the initial slurry pump pipeline flow mapping equation. If the mapping requirements are met, there is no need to adjust the independent variable coefficients and bias. By setting the mapping Euclidean distance threshold, a quantitative judgment basis is provided for determining whether the mapping result of the slurry pump pipeline flow mapping equation meets the requirements.

[0042] Preferably, optimizing and adjusting the initial slurry pump pipeline flow mapping equation in S323 until the mapped Euclidean distance is less than a mapped Euclidean distance threshold comprises the following steps:

[0043] S3231, Constructing flow mapping equations for tuna populations d 1i represents the i-th tuna in the tuna population of the flow mapping equation, d1′ represents the size of the tuna population of the flow mapping equation; the maximum number of iterations of the tuna population of the flow mapping equation is set to And the current iteration number is are respectively recorded as the maximum number of iterations of the flow mapping equation and the current number of iterations of the flow mapping equation; the search space dimension of the tuna population of the flow mapping equation is a′2+b′+1;

[0044] S3232, setting the value intervals of each independent variable coefficient and the offset in the initial slurry pump pipeline flow mapping equation to obtain the independent variable coefficient value interval matrix And the bias value range respectively represent the lower limit and the upper limit of the bias value in the initial slurry pump pipeline flow mapping equation; as follows,

[0045]

[0046] in, They respectively represent the lower limit and upper limit of the value of the i-th independent variable coefficient of the slurry pump pipeline flow mapping equation; They respectively represent the lower limit and upper limit of the (a′2+i)th independent variable coefficient of the slurry pump pipeline flow mapping equation;

[0047] The initial position matrix of each tuna in the flow mapping equation tuna population is set according to the independent variable coefficient value interval matrix and the bias value interval, and the first initial position matrix set e1 = {e 11 ,...,e1k ,...,e 1d1 ′},e 1k The initial position matrix of the i-th tuna in the tuna population in the flow mapping equation is as follows,

[0048]

[0049] Among them, e 1k1i 、e 1k2i They respectively represent the components of the initial position matrix of the i-th tuna in the tuna population in the flow mapping equation in the i-th independent variable coefficient and the (a′2+i)-th independent variable coefficient dimensions of the slurry pump pipeline flow mapping equation; e 1k3 represents the component of the initial position matrix of the i-th tuna in the tuna population in the flow mapping equation in the offset dimension of the slurry pump pipeline flow mapping equation; e 1k1i 、e 1k2i 、e 1k3 The calculation formulas are as follows:

[0050]

[0051] Among them, rand 1k1i 、rand 1k2i 、rand 1k3 Respectively for e 1k1i 、e 1k2i 、e 1k3 Generate a random number between 0 and 1;

[0052] S3233, according to the mapped Euclidean distance Construct the fitness function e′ of the tuna population of the flow mapping equation as follows:

[0053]

[0054] S3234, start iteration, and set the current iteration number of the flow mapping equation to 1 before iteration; in the first round of iteration, use the fitness function e′ of the tuna population of the flow mapping equation to calculate the fitness value of the initial position matrix of each tuna in the first initial position matrix set to obtain a first fitness value set; use the maximum fitness value in the first fitness value set and the corresponding initial position matrix of the tuna as the first global optimal fitness and the first global optimal position respectively; update the initial position matrix of each tuna in the first initial position matrix set according to the first global optimal fitness and the first global optimal position; after the update is completed, add 1 to the current iteration number of the flow mapping equation and enter the next round of iteration;

[0055] In each other round of iteration, the fitness function e′ of the tuna population of the flow mapping equation is used to calculate the fitness value of the position matrix of each tuna in the tuna population of the flow mapping equation updated in the previous round of iteration to obtain a second fitness value set; the maximum fitness value in the second fitness value set and the corresponding tuna position matrix are respectively used as the second global optimal fitness and the second global optimal position; the position matrix of each tuna in the tuna population of the flow mapping equation updated in the previous round of iteration is updated according to the second global optimal fitness and the second global optimal position; after the update is completed, the current iteration number of the flow mapping equation is increased by 1 and the next iteration is entered;

[0056] S3235, when When , stop the iteration and obtain the first final global optimal position; substitute each position component of the first final global optimal position into the initial slurry pump pipeline flow mapping equation respectively to obtain the optimized slurry pump pipeline flow mapping equation; combine the current monitoring data matrix set and each monitoring data set in the current monitoring image feature data matrix set and the corresponding monitoring image feature data set and substitute them into the optimized slurry pump pipeline flow mapping equation for mapping to obtain the current optimized flow mapping data matrix; calculate the Euclidean distance between the current optimized flow mapping data matrix and the current flow data matrix to obtain the optimized mapping Euclidean distance;

[0057] When the optimized mapped Euclidean distance is less than the mapped Euclidean distance threshold, the optimized slurry pump pipeline flow mapping equation is used as the final slurry pump pipeline flow mapping equation; otherwise, return to S3234 to continue iterating until the optimized mapped Euclidean distance is less than the mapped Euclidean distance threshold;

[0058] The tuna optimization algorithm can find a more accurate optimal solution in the search space; through the two strategies of spiral foraging and parabolic foraging, it can explore globally during the search process to avoid falling into the local optimal solution; based on the above advantages, the tuna optimization algorithm is used in this scheme to iteratively update the coefficients of each independent variable and the offset of the initial slurry pump pipeline flow mapping equation for multiple times, and the Euclidean distance between the mapping data of the slurry pump pipeline flow mapping equation and the actual data is used as the fitness function; therefore, as the iteration proceeds, after the obtained slurry pump pipeline flow mapping equation is substituted into the initial slurry pump pipeline flow mapping equation, the mapping accuracy of the initial slurry pump pipeline flow mapping equation becomes higher and higher, and finally meets the mapping requirements.

[0059] Preferably, S4 comprises the following steps:

[0060] S41, constructing an initial pipeline data prediction model; using the current monitoring data matrix set to train and test the initial pipeline flow prediction model to obtain a final pipeline data prediction model;

[0061] Constructing an initial pipeline image feature data prediction model; using the current monitoring image feature data matrix set to train and test the initial pipeline image feature data prediction model to obtain a final pipeline image feature data prediction model;

[0062] S42, set the number of real-time data collection According to the number of real-time collected data, sensors are arranged at each monitoring node in the monitoring node set to collect data and image data in the pipeline of the slurry pump to be monitored, and a matrix set of monitoring data to be predicted is obtained. and the matrix of the monitoring image data to be predicted; and then using the fluid image feature data extraction model to extract the particle distribution feature data of each image data in the matrix of the monitoring image data to be predicted, to obtain the matrix set of the monitoring image feature data to be predicted They respectively represent the monitoring data matrix and the monitoring image feature data matrix at the i-th time point collected according to the number of real-time collected data; they are as follows,

[0063]

[0064] in, express The data collected by the k-th type of data monitoring sensor at the j-th monitoring node; express The k-th type of particle distribution characteristic data of the image data collected at the j-th monitoring node;

[0065] S43, setting the number of future predictions f; inputting the to-be-predicted monitoring data matrix set and the to-be-predicted monitoring image feature data matrix set into the final pipeline data prediction model and the final pipeline image feature data prediction model for prediction according to the future prediction number, and obtaining the future monitoring data matrix set f1′={f1′1,...,f1′ i ,...,f1′ f} and the future monitoring image feature data matrix set f2′={f2′1,...,f2′ i ,...,f2′ f}; f1′ i 、f2′ i They represent the monitoring data matrix and monitoring image feature data matrix predicted at the i-th future time point respectively; they are as follows,

[0066]

[0067] Among them, f1′ ijk 、f2′ ijk Respectively represent f1′ i The kth type of monitoring data at the jth monitoring node and f2′ i The k-th type of particle distribution characteristic data of the image data at the j-th monitoring node;

[0068] The initial pipeline data prediction model and the final pipeline image feature data prediction model are preferably BP neural network models;

[0069] It provides a prediction model for building the final pipeline data prediction model and the final pipeline image feature data prediction model, and provides a prediction model for the subsequent prediction of various data of the fluid in the monitored slurry pump pipeline collected in real time and the particle distribution adjustment data, so as to timely predict the abnormal flow conditions that may occur in the slurry pump pipeline, and then take corresponding measures to control the flow.

[0070] Preferably, the S5 comprises the following steps:

[0071] S51, inputting the future monitoring data matrix set and the future monitoring image feature data matrix set into the final slurry pump pipeline flow mapping equation for mapping, and obtaining the future pipeline flow data matrix as follows,

[0072]

[0073] in, represents the pipeline flow data at the jth monitoring node at the i-th future time point predicted;

[0074] Accumulate each row of data in the future pipeline flow data matrix to obtain the future pipeline overall flow data set Represents the overall flow data of the pipeline at the i-th future time point; the calculation formula is as follows,

[0075]

[0076] S52, setting a flow rate standard interval range; when each future pipeline overall flow rate data in the future pipeline overall flow rate data set is within the flow rate standard interval range, there is no need to adjust the slurry pump motor frequency; otherwise, the slurry pump motor frequency is adjusted, and S42, S43 and S51 are repeated until each future pipeline overall flow rate data in the future pipeline overall flow rate data set is within the flow rate standard interval range;

[0077] Providing a set flow standard interval range provides a quantitative basis for determining whether the flow data in the monitored slurry pump pipeline is normal.

[0078] A slurry pump pipeline flow adaptive monitoring system based on machine learning, comprising a sensor type setting module, a slurry pump pipeline fluid data acquisition module to be monitored, a historical pipeline fluid data acquisition module, a fluid image feature data extraction model construction module, a current image feature extraction module, a flow mapping equation construction module, a prediction model construction module, a real-time slurry pump pipeline fluid data acquisition module, a prediction module, a mapping module and a flow adjustment module;

[0079] The sensor type setting module is used to set a variety of sensor types for collecting data and sensor types for collecting images for monitoring flow, and obtain a data monitoring sensor type set;

[0080] The monitored slurry pump pipeline fluid data acquisition module is used to collect data in the monitored slurry pump pipeline at multiple time points according to the data monitoring sensor type set to obtain a current monitoring data matrix set, a current monitoring image data matrix and a current flow data matrix;

[0081] The historical pipeline fluid data acquisition module is used to collect multiple historical slurry pump pipeline fluid image data and corresponding particle distribution feature data according to the data monitoring sensor type set, and obtain the historical fluid original image data set and the historical fluid image feature data matrix;

[0082] The fluid image feature data extraction model construction module is used to construct a fluid image feature data extraction model using a historical fluid original image data set and a historical fluid image feature data matrix;

[0083] The current image feature extraction module is used to extract particle distribution features from each monitoring image data in the current monitoring image data matrix using a fluid image feature data extraction model to obtain a current monitoring image feature data matrix set;

[0084] The flow mapping equation construction module is used to construct the final slurry pump pipeline flow mapping equation using the current monitoring image feature data matrix set, the current monitoring data matrix set and the current flow data matrix;

[0085] The prediction model building module is used to respectively use the current monitoring data matrix set and the current monitoring image feature data matrix set to build a final pipeline data prediction model and a final pipeline image feature data prediction model;

[0086] The real-time slurry pump pipeline fluid data acquisition module is used to collect data in the slurry pump pipeline to be monitored in real time for multiple times and use the fluid image feature data extraction model to extract the particle distribution characteristics of the image data therein to obtain a matrix set of monitoring data to be predicted and a matrix set of monitoring image feature data to be predicted;

[0087] The prediction module is used to predict the slurry pump pipeline fluid data at future moments according to the monitoring data matrix set to be predicted and the monitoring image feature data matrix set to be predicted, respectively using the final pipeline data prediction model and the final pipeline image feature data prediction model to obtain the future monitoring data matrix set and the future monitoring image feature data matrix set;

[0088] The mapping module is used to map and calculate the future monitoring data matrix set and the future monitoring image feature data matrix set using the final slurry pump pipeline flow mapping equation to obtain the future pipeline overall flow data set;

[0089] The flow adjustment module is used to adjust the frequency of the slurry pump motor according to the future pipeline overall flow data set to adjust the flow in the slurry pump pipeline.

[0090] The present invention has the following beneficial effects:

[0091] 1. The present invention monitors and maps the data in the slurry pump pipeline from various types of data such as concentration, particle size distribution, pressure, temperature, etc., and considers various factors to make the mapping more accurate and comprehensive; by constructing the final slurry pump pipeline flow mapping equation, the fluid data of multiple slurry pump pipelines to be monitored that are subsequently collected in real time are mapped, so as to obtain the corresponding flow data, and further provide data support for the final determination of whether the flow data of the slurry pump pipeline to be monitored meets the requirements; by predicting the slurry pump pipeline fluid data at future times and obtaining the corresponding flow data, it is possible to determine whether there is a problem with the current motor frequency of the slurry pump to be monitored. If there is a problem, it can be adjusted in advance to ensure that the slurry pump can work normally and stably.

[0092] 2. In the present invention, each image data in the current monitoring image data matrix is ​​input into the fluid image feature data extraction model to extract its particle distribution feature data, thereby providing support for particle distribution parameter data for the subsequent construction of the final slurry pump pipeline flow mapping equation, thereby facilitating the construction of a mapping equation related to the particle distribution parameter data.

[0093] 3. The present invention adopts the tuna optimization algorithm to simultaneously iteratively update the coefficients of each independent variable and the offset of the initial slurry pump pipeline flow mapping equation for multiple times, and uses the Euclidean distance between the mapping data of the slurry pump pipeline flow mapping equation and the actual data as the fitness function; therefore, as the iteration proceeds, the mapping accuracy of the initial slurry pump pipeline flow mapping equation becomes higher and higher, and finally meets the mapping requirements.

[0094] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0096] Figure 1 It is a flow chart of the method for adaptively monitoring flow in a slurry pump pipeline based on machine learning of the present invention. DETAILED DESCRIPTION

[0097] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0098] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0099] Embodiment 1

[0100] See also Figure 1 , this embodiment is a method for adaptively monitoring the flow rate of a slurry pump pipeline based on machine learning, comprising the following steps:

[0101] S1, collecting data in the pipeline of the slurry pump to be monitored at multiple time points to obtain a current monitoring data matrix set, a current monitoring image data matrix and a current flow data matrix;

[0102] The S1 comprises the following steps:

[0103] S11, setting a slurry pump pipeline to be monitored; evenly setting a number of monitoring nodes in the slurry pump pipeline to be monitored to obtain a monitoring node set a 1i represents the i-th monitoring node set in the pipeline of the slurry pump to be monitored, and a′1 represents the total number of monitoring nodes set in the pipeline of the slurry pump to be monitored; set a variety of sensor types for collecting data for monitoring flow and sensor types for collecting images, and obtain a data monitoring sensor type set and image monitoring sensor type, a 2irepresents the setting of the i-th data monitoring sensor type, a′2 represents the total number of the set data monitoring sensor types;

[0104] S12, evenly set multiple data collection time points to obtain a real-time data collection time point set; arrange sensors at each monitoring node in the monitoring node set according to the data monitoring sensor type set, the image monitoring sensor type and the real-time data collection time point set, and collect data in the pipeline of the slurry pump to be monitored after the slurry pump starts working, to obtain a current monitoring data matrix set Current monitoring image data matrix And the current traffic data matrix represents the data matrix monitored at the i-th time point in the data real-time acquisition time point set, Indicates the total number of set data collection time points; They are as follows:

[0105]

[0106] in, express The data collected by the k-th type of data monitoring sensor at the j-th monitoring node, represents the image data collected at the jth monitoring node at the i-th data collection time point, represents the flow data collected at the jth monitoring node at the i-th data collection time point;

[0107] S2, collecting multiple historical fluid image data in the slurry pump pipeline and corresponding particle distribution feature data in cooperation with the current monitoring image data matrix and constructing a fluid image feature data extraction model;

[0108] The S2 comprises the following steps:

[0109] S21, setting multiple particle distribution feature types of the fluid image in the slurry pump pipeline, obtaining a fluid image feature type set b = {b1, ..., b i ,...,b b′}, b i represents the i-th particle distribution feature type of the set fluid image in the slurry pump pipeline, and b′ represents the total number of particle distribution feature types of the set fluid image in the slurry pump pipeline;

[0110] According to the fluid image feature type set, multiple fluid image data in the slurry pump pipeline and corresponding particle distribution feature data are collected to obtain a historical fluid original image data set. And the historical fluid image feature data matrix represents the i-th historical fluid original image data collected, Indicates the total amount of historical fluid raw image data collected; as follows,

[0111]

[0112] in, Represents the j-th type of particle distribution characteristic data of the i-th historical fluid original image data collected;

[0113] S22, constructing a fluid image feature data extraction model using the historical fluid original image data set and the historical fluid image feature data matrix;

[0114] The S22 comprises the following steps:

[0115] S221, constructing an initial CNN neural network model and setting a first training data ratio; dividing the historical fluid original image data set and the historical fluid image feature data matrix according to the first training data ratio to obtain a historical fluid original image training data set, a historical fluid image feature training data matrix, a historical fluid original image test data set, and a historical fluid image feature test data matrix;

[0116] S222, setting a first training error threshold; annotating each original image training data in the historical fluid original image training data set according to the historical fluid image feature training data matrix to obtain a historical fluid annotated image training data set; inputting the historical fluid annotated image training data set as training data into the initial CNN neural network model for training; during the training process, when the training error is less than the first training error threshold, stopping the training to obtain a trained CNN neural network model; otherwise, continuing the training until the training error is less than the first training error threshold;

[0117] S223, setting a first test accuracy threshold; inputting the historical fluid original image test data set and the historical fluid image feature test data matrix as test data and test labels respectively into the trained CNN neural network model for testing; after the test is completed, obtaining the first test accuracy data; when the first test accuracy data is greater than or equal to the first test accuracy threshold, using the trained CNN neural network model as the fluid image feature data extraction model; otherwise, returning to S222 to continue training the trained CNN neural network model until the first test accuracy data is greater than or equal to the first test accuracy threshold;

[0118] S3, using the fluid image feature data extraction model to extract particle distribution features from each monitoring image data in the current monitoring image data matrix, and constructing the final slurry pump pipeline flow mapping equation in conjunction with the current monitoring data matrix set and the current flow data matrix;

[0119] The S3 comprises the following steps:

[0120] S31, input each image data in the current monitoring image data matrix into the fluid image feature data extraction model to extract its particle distribution feature data, and obtain the current monitoring image feature data matrix set It represents the particle distribution characteristic data matrix of the fluid in the pipeline of the slurry pump to be monitored at the i-th time point in the real-time data collection time point set; as follows,

[0121]

[0122] in, express The k-th type of particle distribution characteristic data of the image data collected at the j-th monitoring node;

[0123] S32, constructing a final slurry pump pipeline flow mapping equation using the current monitoring data matrix set, the current monitoring image feature data matrix set and the current flow data matrix;

[0124] The S32 comprises the following steps:

[0125] S321, constructing the initial slurry pump pipeline flow mapping equation; as follows,

[0126]

[0127] Wherein, c is the dependent variable of the initial slurry pump pipeline flow mapping equation, which represents the slurry pump pipeline flow data; c′ 2i , c′ 4i are the ith independent variable and the (a′2+i)th independent variable of the slurry pump pipeline flow mapping equation, respectively, representing the data collected by the ith type of data monitoring sensor and the ith type of particle distribution characteristic data, c1′ i 、c3′ i c′ 2i , c′ 4i The independent variable coefficient; c5′ represents the offset of the initial slurry pump pipeline flow mapping equation;

[0128] S322, each monitoring data set in the current monitoring data matrix set and the current monitoring image feature data matrix set and the corresponding monitoring image feature data set are combined and substituted into the initial slurry pump pipeline flow mapping equation for mapping, and the current flow mapping data matrix is ​​obtained. as follows,

[0129]

[0130] in, Indicates that The jth row of data in Substitute the j-th row data in into the initial slurry pump pipeline flow mapping equation to obtain the flow mapping data;

[0131] S323, setting a mapping Euclidean distance threshold; calculating the Euclidean distance between the current traffic mapping data matrix and the current traffic data matrix to obtain a mapping Euclidean distance c~; the calculation formula is as follows,

[0132]

[0133] When the mapped Euclidean distance is less than the mapped Euclidean distance threshold, there is no need to optimize and adjust the initial slurry pump pipeline flow mapping equation; otherwise, the initial slurry pump pipeline flow mapping equation is optimized and adjusted until the mapped Euclidean distance is less than the mapped Euclidean distance threshold;

[0134] S323 optimizes and adjusts the initial slurry pump pipeline flow mapping equation until the mapped Euclidean distance is less than the mapped Euclidean distance threshold, including the following steps:

[0135] S3231, construct flow mapping equation tuna population d1 = {d 11 ,...,d 1i ,...,d 1d1′}, d 1i represents the i-th tuna in the tuna population of the flow mapping equation, d1′ represents the size of the tuna population of the flow mapping equation; the maximum number of iterations of the tuna population of the flow mapping equation is set to And the current iteration number is are respectively recorded as the maximum number of iterations of the flow mapping equation and the current number of iterations of the flow mapping equation; the search space dimension of the tuna population of the flow mapping equation is a′2+b′+1;

[0136] S3232, setting the value intervals of each independent variable coefficient and the offset in the initial slurry pump pipeline flow mapping equation to obtain the independent variable coefficient value interval matrix And the bias value range respectively represent the lower limit and the upper limit of the bias value in the initial slurry pump pipeline flow mapping equation; as follows,

[0137]

[0138] in, They respectively represent the lower limit and upper limit of the value of the i-th independent variable coefficient of the slurry pump pipeline flow mapping equation; They respectively represent the lower limit and upper limit of the (a′2+i)th independent variable coefficient of the slurry pump pipeline flow mapping equation;

[0139] The initial position matrix of each tuna in the flow mapping equation tuna population is set according to the independent variable coefficient value interval matrix and the bias value interval, and the first initial position matrix set e1 = {e 11 ,...,e 1k ,...,e 1d1 ′},e 1k The initial position matrix of the i-th tuna in the tuna population in the flow mapping equation is as follows,

[0140]

[0141] Among them, e 1k1i 、e 1k2i They respectively represent the components of the initial position matrix of the i-th tuna in the tuna population in the flow mapping equation in the i-th independent variable coefficient and the (a′2+i)-th independent variable coefficient dimensions of the slurry pump pipeline flow mapping equation; e 1k3 represents the component of the initial position matrix of the i-th tuna in the tuna population in the flow mapping equation in the offset dimension of the slurry pump pipeline flow mapping equation; e 1k1i 、e 1k2i 、e 1k3 The calculation formulas are as follows:

[0142]

[0143] Among them, rand 1k1i 、rand 1k2i 、rand 1k3 Respectively for e 1k1i 、e 1k2i 、e 1k3 Generate a random number between 0 and 1;

[0144] S3233, according to the mapped Euclidean distance Construct the fitness function e′ of the tuna population of the flow mapping equation as follows:

[0145]

[0146] S3234, start iteration, and set the current iteration number of the flow mapping equation to 1 before iteration; in the first round of iteration, use the fitness function e′ of the tuna population of the flow mapping equation to calculate the fitness value of the initial position matrix of each tuna in the first initial position matrix set to obtain a first fitness value set; use the maximum fitness value in the first fitness value set and the corresponding initial position matrix of the tuna as the first global optimal fitness and the first global optimal position respectively; update the initial position matrix of each tuna in the first initial position matrix set according to the first global optimal fitness and the first global optimal position; after the update is completed, add 1 to the current iteration number of the flow mapping equation and enter the next round of iteration;

[0147] In each other round of iteration, the fitness function e′ of the tuna population of the flow mapping equation is used to calculate the fitness value of the position matrix of each tuna in the tuna population of the flow mapping equation updated in the previous round of iteration to obtain a second fitness value set; the maximum fitness value in the second fitness value set and the corresponding tuna position matrix are respectively used as the second global optimal fitness and the second global optimal position; the position matrix of each tuna in the tuna population of the flow mapping equation updated in the previous round of iteration is updated according to the second global optimal fitness and the second global optimal position; after the update is completed, the current iteration number of the flow mapping equation is increased by 1 and the next iteration is entered;

[0148] S3235, when When , stop the iteration and obtain the first final global optimal position; substitute each position component of the first final global optimal position into the initial slurry pump pipeline flow mapping equation respectively to obtain the optimized slurry pump pipeline flow mapping equation; combine the current monitoring data matrix set and each monitoring data set in the current monitoring image feature data matrix set and the corresponding monitoring image feature data set and substitute them into the optimized slurry pump pipeline flow mapping equation for mapping to obtain the current optimized flow mapping data matrix; calculate the Euclidean distance between the current optimized flow mapping data matrix and the current flow data matrix to obtain the optimized mapping Euclidean distance;

[0149] When the optimized mapped Euclidean distance is less than the mapped Euclidean distance threshold, the optimized slurry pump pipeline flow mapping equation is used as the final slurry pump pipeline flow mapping equation; otherwise, return to S3234 to continue iterating until the optimized mapped Euclidean distance is less than the mapped Euclidean distance threshold;

[0150] S4, collecting data in the pipeline of the slurry pump to be monitored in real time for multiple times and extracting the particle distribution characteristics of the image data using a fluid image feature data extraction model, and predicting the fluid data of the slurry pump pipeline at future times with the extracted particle distribution feature data to obtain a future monitoring data matrix set and a future monitoring image feature data matrix set;

[0151] The S4 comprises the following steps:

[0152] S41, constructing an initial pipeline data prediction model; using the current monitoring data matrix set to train and test the initial pipeline flow prediction model to obtain a final pipeline data prediction model;

[0153] Constructing an initial pipeline image feature data prediction model; using the current monitoring image feature data matrix set to train and test the initial pipeline image feature data prediction model to obtain a final pipeline image feature data prediction model;

[0154] S42, set the number of real-time data collection According to the number of real-time collected data, sensors are arranged at each monitoring node in the monitoring node set to collect data and image data in the pipeline of the slurry pump to be monitored, and a matrix set of monitoring data to be predicted is obtained. and the matrix of the monitoring image data to be predicted; and then using the fluid image feature data extraction model to extract the particle distribution feature data of each image data in the matrix of the monitoring image data to be predicted, to obtain the matrix set of the monitoring image feature data to be predicted They respectively represent the monitoring data matrix and the monitoring image feature data matrix at the i-th time point collected according to the number of real-time collected data; they are as follows,

[0155]

[0156] in, express The data collected by the k-th type of data monitoring sensor at the j-th monitoring node; express The k-th type of particle distribution characteristic data of the image data collected at the j-th monitoring node;

[0157] S43, setting the number of future predictions f; inputting the to-be-predicted monitoring data matrix set and the to-be-predicted monitoring image feature data matrix set into the final pipeline data prediction model and the final pipeline image feature data prediction model for prediction according to the future prediction number, and obtaining the future monitoring data matrix set f1′={f1′1,...,f1′ i ,...,f1′ f} and the future monitoring image feature data matrix set f2′={f2′1,...,f2′ i ,...,f2′ f}; f1′ i 、f2′ i They represent the monitoring data matrix and monitoring image feature data matrix predicted at the i-th future time point respectively; they are as follows,

[0158]

[0159] Among them, f1′ ijk 、f2′ ijk Respectively represent f1′ i The kth type of monitoring data at the jth monitoring node and f2′ i The k-th type of particle distribution characteristic data of the image data at the j-th monitoring node;

[0160] S5. Use the final slurry pump pipeline flow mapping equation to map and calculate the future monitoring data matrix set and the future monitoring image feature data matrix set to obtain the future pipeline overall flow data set; adjust the frequency of the slurry pump motor according to the future pipeline overall flow data set to adjust the flow in the slurry pump pipeline;

[0161] The S5 comprises the following steps:

[0162] S51, inputting the future monitoring data matrix set and the future monitoring image feature data matrix set into the final slurry pump pipeline flow mapping equation for mapping, and obtaining the future pipeline flow data matrix as follows,

[0163]

[0164] in, represents the pipeline flow data at the jth monitoring node at the i-th future time point predicted;

[0165] Accumulate each row of data in the future pipeline flow data matrix to obtain the future pipeline overall flow data set Represents the overall flow data of the pipeline at the i-th future time point; the calculation formula is as follows,

[0166]

[0167] S52, setting the flow standard interval range; when each future pipeline overall flow data in the future pipeline overall flow data set is within the flow standard interval range, there is no need to adjust the slurry pump motor frequency; otherwise, adjust the slurry pump motor frequency, and repeat S42, S43 and S51 until each future pipeline overall flow data in the future pipeline overall flow data set is within the flow standard interval range.

[0168] Embodiment 2

[0169] The present embodiment discloses a slurry pump pipeline flow adaptive monitoring system based on machine learning, the system can implement the method of the above embodiment, including a sensor type setting module, a slurry pump pipeline fluid data acquisition module to be monitored, a historical pipeline fluid data acquisition module, a fluid image feature data extraction model construction module, a current image feature extraction module, a flow mapping equation construction module, a prediction model construction module, a real-time slurry pump pipeline fluid data acquisition module, a prediction module, a mapping module and a flow adjustment module;

[0170] The sensor type setting module is used to set a variety of sensor types for collecting data and sensor types for collecting images for monitoring flow, and obtain a data monitoring sensor type set;

[0171] The monitored slurry pump pipeline fluid data acquisition module is used to collect data in the monitored slurry pump pipeline at multiple time points according to the data monitoring sensor type set to obtain a current monitoring data matrix set, a current monitoring image data matrix and a current flow data matrix;

[0172] The historical pipeline fluid data acquisition module is used to collect multiple historical slurry pump pipeline fluid image data and corresponding particle distribution feature data according to the data monitoring sensor type set, and obtain the historical fluid original image data set and the historical fluid image feature data matrix;

[0173] The fluid image feature data extraction model construction module is used to construct a fluid image feature data extraction model using a historical fluid original image data set and a historical fluid image feature data matrix;

[0174] The current image feature extraction module is used to extract particle distribution features from each monitoring image data in the current monitoring image data matrix using a fluid image feature data extraction model to obtain a current monitoring image feature data matrix set;

[0175] The flow mapping equation construction module is used to construct the final slurry pump pipeline flow mapping equation using the current monitoring image feature data matrix set, the current monitoring data matrix set and the current flow data matrix;

[0176] The prediction model building module is used to respectively use the current monitoring data matrix set and the current monitoring image feature data matrix set to build a final pipeline data prediction model and a final pipeline image feature data prediction model;

[0177] The real-time slurry pump pipeline fluid data acquisition module is used to collect data in the slurry pump pipeline to be monitored in real time for multiple times and use the fluid image feature data extraction model to extract the particle distribution characteristics of the image data therein to obtain a matrix set of monitoring data to be predicted and a matrix set of monitoring image feature data to be predicted;

[0178] The prediction module is used to predict the slurry pump pipeline fluid data at future moments according to the monitoring data matrix set to be predicted and the monitoring image feature data matrix set to be predicted, respectively using the final pipeline data prediction model and the final pipeline image feature data prediction model to obtain the future monitoring data matrix set and the future monitoring image feature data matrix set;

[0179] The mapping module is used to map and calculate the future monitoring data matrix set and the future monitoring image feature data matrix set using the final slurry pump pipeline flow mapping equation to obtain the future pipeline overall flow data set;

[0180] The flow adjustment module is used to adjust the frequency of the slurry pump motor according to the future pipeline overall flow data set to adjust the flow in the slurry pump pipeline.

[0181] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0182] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A method for adaptively monitoring flow in a slurry pump pipeline based on machine learning, characterized in that: The following steps are involved: S1, collecting data in the pipeline of the slurry pump to be monitored at multiple time points to obtain a current monitoring data matrix set, a current monitoring image data matrix and a current flow data matrix; S2, collecting multiple historical fluid image data in the slurry pump pipeline and corresponding particle distribution feature data in cooperation with the current monitoring image data matrix and constructing a fluid image feature data extraction model; S3, using the fluid image feature data extraction model to extract particle distribution features from each monitoring image data in the current monitoring image data matrix, and constructing the final slurry pump pipeline flow mapping equation in conjunction with the current monitoring data matrix set and the current flow data matrix; S4, collecting data in the pipeline of the slurry pump to be monitored in real time for multiple times and extracting the particle distribution characteristics of the image data using a fluid image feature data extraction model, and predicting the fluid data of the slurry pump pipeline at future times with the extracted particle distribution feature data to obtain a future monitoring data matrix set and a future monitoring image feature data matrix set; S5. Use the final slurry pump pipeline flow mapping equation to map and calculate the future monitoring data matrix set and the future monitoring image feature data matrix set to obtain the future pipeline overall flow data set; adjust the frequency of the slurry pump motor according to the future pipeline overall flow data set to adjust the flow in the slurry pump pipeline.

2. The method for adaptively monitoring flow in a slurry pump pipeline based on machine learning according to claim 1 is characterized in that: The S1 comprises the following steps: S11, setting a slurry pump pipeline to be monitored; evenly setting a number of monitoring nodes in the slurry pump pipeline to be monitored to obtain a monitoring node set; setting a variety of sensor types for collecting data and sensor types for collecting images for monitoring flow, to obtain a data monitoring sensor type set and an image monitoring sensor type; S12. Arrange sensors at each monitoring node in the monitoring node set according to the data monitoring sensor type set and the image monitoring sensor type data, and collect data in the slurry pump pipeline to be monitored after the slurry pump starts working, to obtain a current monitoring data matrix set, a current monitoring image data matrix, and a current flow data matrix.

3. The method for adaptively monitoring slurry pump pipeline flow based on machine learning according to claim 2 is characterized in that: The S2 comprises the following steps: S21, setting a plurality of particle distribution feature types of the fluid image in the slurry pump pipeline to obtain a fluid image feature type set; collecting a plurality of fluid image data in the slurry pump pipeline and corresponding particle distribution feature data according to the fluid image feature type set to obtain a historical fluid original image data set and a historical fluid image feature data matrix; S22, constructing a fluid image feature data extraction model using the historical fluid original image data set and the historical fluid image feature data matrix.

4. The method for adaptively monitoring flow in a slurry pump pipeline based on machine learning according to claim 3 is characterized in that: The fluid image feature data extraction model adopts a CNN neural network model.

5. The method for adaptively monitoring slurry pump pipeline flow based on machine learning according to claim 3 is characterized in that: The S3 comprises the following steps: S31, inputting each image data in the current monitoring image data matrix into a fluid image feature data extraction model to extract its particle distribution feature data, thereby obtaining a current monitoring image feature data matrix set; S32, constructing a final slurry pump pipeline flow mapping equation using the current monitoring data matrix set, the current monitoring image feature data matrix set and the current flow data matrix.

6. The method for adaptively monitoring flow in a slurry pump pipeline based on machine learning according to claim 5 is characterized in that: The S32 comprises the following steps: S321, constructing an initial slurry pump pipeline flow mapping equation; S322, combining the current monitoring data matrix set and each monitoring data set in the current monitoring image feature data matrix set and the corresponding monitoring image feature data set and substituting them into the initial slurry pump pipeline flow mapping equation for mapping, to obtain the current flow mapping data matrix; S323, setting a mapping Euclidean distance threshold; calculating the Euclidean distance between the current flow mapping data matrix and the current flow data matrix to obtain a mapping Euclidean distance; when the mapping Euclidean distance is less than the mapping Euclidean distance threshold, there is no need to optimize and adjust the initial slurry pump pipeline flow mapping equation; otherwise, optimize and adjust the initial slurry pump pipeline flow mapping equation until the mapping Euclidean distance is less than the mapping Euclidean distance threshold.

7. The method for adaptively monitoring flow in a slurry pump pipeline based on machine learning according to claim 6 is characterized in that: S323 optimizes and adjusts the initial slurry pump pipeline flow mapping equation until the mapped Euclidean distance is less than the mapped Euclidean distance threshold, including the following steps: S3231, constructing a flow mapping equation for tuna population; setting the maximum number of iterations of the flow mapping equation for tuna population to And the current number of iterations is They are respectively recorded as the maximum number of iterations of the flow mapping equation and the current number of iterations of the flow mapping equation; S3232, setting the value range of each independent variable coefficient and the offset in the initial slurry pump pipeline flow mapping equation and setting the initial position matrix of each tuna in the tuna population of the flow mapping equation to obtain a first initial position matrix set; S3233, constructing a fitness function of the tuna population of the flow mapping equation according to the mapped Euclidean distance; S3234, start iteration; in each round of iteration, use the fitness function of the flow mapping equation tuna population to calculate the fitness value of the position matrix of each tuna in the flow mapping equation tuna population updated in the previous round of iteration, and update the position matrix of each tuna in the flow mapping equation tuna population updated in the previous round of iteration; S3235, when When , stop the iteration and obtain the first final global optimal position; substitute each position component of the first final global optimal position into the initial slurry pump pipeline flow mapping equation respectively to obtain the optimized slurry pump pipeline flow mapping equation; combine the current monitoring data matrix set and each monitoring data set in the current monitoring image feature data matrix set and the corresponding monitoring image feature data set and substitute them into the optimized slurry pump pipeline flow mapping equation for mapping to obtain the current optimized flow mapping data matrix; calculate the Euclidean distance between the current optimized flow mapping data matrix and the current flow data matrix to obtain the optimized mapping Euclidean distance; When the optimized mapping Euclidean distance is less than the mapping Euclidean distance threshold, the optimized slurry pump pipeline flow mapping equation is used as the final slurry pump pipeline flow mapping equation; otherwise, return to S3234 to continue iterating until the optimized mapping Euclidean distance is less than the mapping Euclidean distance threshold.

8. The method for adaptively monitoring flow in a slurry pump pipeline based on machine learning according to claim 7, characterized in that: The S4 comprises the following steps: S41, constructing an initial pipeline data prediction model; using the current monitoring data matrix set to train and test the initial pipeline flow prediction model to obtain a final pipeline data prediction model; Constructing an initial pipeline image feature data prediction model; using the current monitoring image feature data matrix set to train and test the initial pipeline image feature data prediction model to obtain a final pipeline image feature data prediction model; S42, setting the number of real-time data collection; according to the number of real-time data collection and using sensors arranged at each monitoring node in the monitoring node set to collect data and image data in the pipeline of the slurry pump to be monitored, to obtain a matrix set of monitoring data to be predicted and a matrix of monitoring image data to be predicted; then using the fluid image feature data extraction model to extract the particle distribution feature data of each image data in the matrix of monitoring image data to be predicted, to obtain a matrix set of monitoring image feature data to be predicted; S43, setting the number of future predictions; inputting the to-be-predicted monitoring data matrix set and the to-be-predicted monitoring image feature data matrix set into the final pipeline data prediction model and the final pipeline image feature data prediction model respectively for prediction according to the future prediction number, and obtaining the future monitoring data matrix set and the future monitoring image feature data matrix set.

9. The method for adaptively monitoring flow in a slurry pump pipeline based on machine learning according to claim 8, characterized in that: The S5 comprises the following steps: S51, inputting the future monitoring data matrix set and the future monitoring image feature data matrix set into the final slurry pump pipeline flow mapping equation for mapping, to obtain a future pipeline flow data matrix; accumulating each row of data in the future pipeline flow data matrix to obtain a future pipeline overall flow data set; S52, setting the flow standard interval range; when each future pipeline overall flow data in the future pipeline overall flow data set is within the flow standard interval range, there is no need to adjust the slurry pump motor frequency; otherwise, adjust the slurry pump motor frequency, and repeat S42, S43 and S51 until each future pipeline overall flow data in the future pipeline overall flow data set is within the flow standard interval range.

10. A system for implementing the method for adaptively monitoring the flow rate of a slurry pump pipeline based on machine learning as described in any one of claims 1 to 9.

Images