A multi-parameter comprehensive monitoring method and system for wet desulfurization

Abstract

The present invention relates to the technical field of data processing, and particularly to a multi-parameter comprehensive monitoring method and system for wet desulfurization. The method includes the steps of: collecting multi-parameter data in the wet desulfurization process, obtaining the priority degree of each parameter in the initial multi-parameter data according to the characteristic intensity of each parameter in the initial multi-parameter data and the similarity degree of each parameter in the initial multi-parameter data with other parameters, constructing an initial isolation tree for the initial multi-parameter data based on the priority degree of each parameter, adding real-time data into the initial isolation tree, calculating the degree of structural change of the initial isolation tree, and adaptively updating the initial isolation tree according to the degree of structural change; finally, obtaining the anomaly score of each multi-parameter data according to the final isolation tree, and giving an early warning according to the anomaly score result. The anomaly detection result of the present invention is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a multi-parameter comprehensive monitoring method and system for wet flue gas desulfurization. Background Art

[0002] The wet flue gas desulfurization technology is mainly used to control sulfur dioxide emissions from coal-fired power plants and other industrial facilities, and it is one of the current relatively common and effective desulfurization technologies. This technology reduces air pollution and protects the environment by bringing the flue gas containing sulfur dioxide into contact with an alkaline absorbent (usually limestone or lime solvent), converting sulfur dioxide into sulfate, and avoiding the negative impacts caused by environmental problems such as acid rain. Since in the wet flue gas desulfurization system, the process of sulfur dioxide in the flue gas reacting with the absorbent to form sulfate involves multiple complex chemical and physical processes, in order to ensure the desulfurization efficiency and optimize the system performance, it is necessary to monitor the abnormal states of multiple parameters. By detecting the abnormal states of these parameters in real time, the operating conditions can be adjusted to ensure the best desulfurization effect, while reducing the operating cost and extending the equipment life.

[0003] The currently publicly available patent application document with the publication number CN109800900A discloses a method for modularizing and visualizing the isolation forest algorithm, including: (1) modularizing the iforest algorithm; dividing the iforest algorithm into two stages, namely training and prediction. The process of training the iforest is made into an independent module, which is the training iforest algorithm module, and the process of predicting the iforest is made into an independent module, which is the prediction iforest algorithm module. At the same time, an evaluation index is added as a module for evaluating the quality of the iforest algorithm, which is the evaluation iforest algorithm module; (2) visualizing the results of each module of the iforest algorithm; visualizing the outputs corresponding to the three modules in step (1) in sequence. The training iforest algorithm module stores the trained model, and the output of the prediction iforest algorithm module is displayed in the data graph, including the predicted labels, and the evaluation iforest algorithm module displays the evaluated indexes.

[0004] The above isolation forest algorithm is an algorithm for detecting anomalies in multi-parameter data. Its working principle is based on the idea of "isolation". By selecting one parameter at each layer of the isolation tree to achieve the segmentation of data samples, abnormal data is segmented at the upper layer of the isolation tree, widening the difference between abnormal data and normal data. If any parameter is randomly selected at different layers of the isolation tree for segmentation, a parameter with poor discrimination may be placed at the upper layer of the isolation tree, making the distance between abnormal data and normal data in the isolation tree relatively close, thus making the calculation of the abnormal score result inaccurate. Summary of the Invention

[0005] To solve the technical problem that when constructing isolation trees by the isolation forest algorithm, any parameter is randomly selected at different layers for splitting, which may place a parameter with poor discrimination degree on the upper layer of the isolation tree, resulting in a short distance between abnormal data and normal data in the isolation tree and inaccurate abnormal score results, the present invention provides a multi-parameter comprehensive monitoring method and system for wet flue gas desulfurization.

[0006] In the first aspect, the present invention provides a multi-parameter comprehensive monitoring method for wet flue gas desulfurization, adopting the following technical solution:

[0007] A multi-parameter comprehensive monitoring method for wet flue gas desulfurization includes the steps of:

[0008] Collect initial multi-parameter data; obtain the priority degree of each parameter in the initial multi-parameter data , represents the feature intensity of the i-th parameter in the initial multi-parameter data; represents the priority degree of the i-th parameter in the initial multi-parameter data; represents the similarity degree between the i-th parameter and other parameters in the initial multi-parameter data; norm() represents the normalization function, and construct an initial isolation tree for the initial multi-parameter data based on the priority degree of each parameter;

[0009] Add real-time data to the initial isolation tree to obtain the first updated isolation tree; obtain the degree of structural change of the first updated isolation tree ; represents the number of layers of the first updated isolation tree; represents the proportion of the number of data of the a-th node in the m-th layer of the first updated isolation tree; represents the proportion of the number of data of the a-th node in the m-th layer of the initial isolation tree; represents the layer number sequence number of the first updated isolation tree; represents the number of nodes in the m-th layer of the first updated isolation tree; exp() represents the exponential function with the natural constant as the base; || represents the absolute value symbol; adaptively update the initial isolation tree based on the degree of structural change of the first updated isolation tree to obtain the final isolation tree; perform abnormal monitoring according to the final isolation tree.

[0010] The innovation of the present invention lies in obtaining the preference degree of each parameter of the initial multi-parameter data according to the change amplitude of each parameter of the initial multi-parameter data, and then constructing an initial isolation tree in the order from large to small of the preference degree, so that a parameter with a larger preference degree can be placed in the upper layer of the isolation tree, enabling the abnormal data and the normal data to be closer in the isolation tree, thereby making the calculation of the abnormal score result accurate and improving the abnormal monitoring effect; and adding the real-time data into the initial isolation tree to obtain an updated isolation tree, and judging whether it is necessary to update the initial isolation tree according to the comparison between the initial isolation tree and the updated isolation tree to obtain the degree of structural change of the updated isolation tree for abnormal monitoring, which can update the isolation tree in real time and further make the abnormal monitoring result more accurate.

[0011] Preferably, the obtaining of the feature strength of the i-th parameter in the initial multi-parameter data includes:

[0012] ;

[0013] In the formula, represents the feature strength of the i-th parameter in the initial multi-parameter data; represents the number of the i-th parameter in the initial multi-parameter data; represents the value of the j-th i-th parameter in the initial multi-parameter data; represents all the i-th parameters in the initial multi-parameter data; norm() represents the normalization function.

[0014] It reflects the change amplitude of each parameter. Subsequently, according to the feature strength of each parameter, the preference degree of each parameter in the initial multi-parameter data is obtained.

[0015] Preferably, the similarity degree between the i-th parameter and other parameters in the initial multi-parameter data includes:

[0016] Obtain the absolute value of the Pearson correlation coefficient between the i-th parameter in the initial multi-parameter data and each parameter except the i-th parameter, and record the mean value of the absolute values of the Pearson correlation coefficients between the i-th parameter in the initial multi-parameter data and all parameters except the i-th parameter as the similarity degree between the i-th parameter and other parameters in the initial multi-parameter data.

[0017] It is convenient to obtain the preference degree of each parameter in the initial multi-parameter data according to the similarity degree between each parameter and other parameters in the initial multi-parameter data.

[0018] Preferably, constructing an initial isolation tree for the initial multi-parameter data based on the preference degree of each parameter includes:

[0019] Construct an initial isolation tree from top to bottom for the initial multi-parameter data in the order of the priority of each parameter in the initial multi-parameter data from largest to smallest.

[0020] The constructed isolation tree can make the abnormal monitoring results more accurate.

[0021] Preferably, adding the real-time data to the initial isolation tree to obtain the first updated isolation tree includes:

[0022] Preset the number of iteratively updated data M. Denote the data corresponding to the Xth sampling moment in the multi-parameter data sequence as the first node in the multi-parameter data sequence, and add the M data after the first node to the initial isolation tree to obtain the first updated isolation tree.

[0023] Preferably, obtaining the final isolation tree includes:

[0024] Preset the number of iteratively updated data M and the change threshold T1. If the structural change degree of the first updated isolation tree is greater than or equal to T1, reconstruct the isolation tree for all the data in the first updated isolation tree according to the construction method of the initial isolation tree to obtain the first final isolation tree; otherwise, denote the first updated isolation tree as the first final isolation tree. Denote the Mth data after the first node in the multi-parameter data sequence as the second node, and add the M data after the second node to the first final isolation tree to obtain the second updated isolation tree. Obtain the structural change degree of the second updated isolation tree to determine whether the second updated isolation tree needs to be reconstructed. And so on, until there is no data in the multi-parameter data sequence, and denote the pth final isolation tree as the final isolation tree.

[0025] Judging whether the initial isolation tree needs to be updated according to the structural change degree of the updated isolation tree and then performing abnormal monitoring can update the isolation tree in real time, and further make the abnormal monitoring results more accurate.

[0026] Preferably, the abnormal monitoring according to the final isolation tree includes:

[0027] Preset the abnormal threshold T2. According to the final isolation tree, obtain the abnormal score of each multi-parameter data. If the mean value of the abnormal scores of all multi-parameter data is greater than the preset abnormal threshold T2, the system issues a warning.

[0028] In a second aspect, the present invention provides a multi-parameter comprehensive monitoring system for wet desulfurization, adopting the following technical solution:

[0029] A multi-parameter comprehensive monitoring system for wet desulfurization includes: a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned multi-parameter comprehensive monitoring method for wet desulfurization is implemented.

[0030] By adopting the above technical solution, a computer program is generated from the above multi-parameter comprehensive monitoring method for wet desulfurization and stored in a memory to be loaded and executed by a processor, so as to manufacture a terminal device according to the memory and the processor, which is convenient to use.

[0031] The present invention has the following technical effects: The purpose of the present invention is to obtain the preference degree of each parameter of the initial multi-parameter data, and then construct an initial isolation tree according to the order of the preference degree from large to small, so that a parameter with a larger preference degree can be placed on the upper layer of the isolation tree, enabling the abnormal data and the normal data to be closer in the isolation tree, thus making the calculation result of the abnormal score accurate and improving the abnormal monitoring effect; and adding real-time data to the initial isolation tree to obtain an updated isolation tree, and judging whether it is necessary to update the initial isolation tree according to the comparison between the initial isolation tree and the updated isolation tree to perform abnormal monitoring, which can update the isolation tree in real time and further make the abnormal monitoring result more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown by way of illustration and not limitation, and the same or corresponding reference numerals denote the same or corresponding parts.

[0033] Figure 1 It is a flowchart of the method in an embodiment of the multi-parameter comprehensive monitoring method for wet desulfurization of the present invention.

[0034] Figure 2 It represents a schematic diagram of the proportion of the number of data. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0036] It should be understood that when the claims, the specification and the drawings of the present invention use the terms "first", "second", etc., they are only used to distinguish different objects and not to describe a specific order. The terms "comprising" and "including" used in the specification and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0037] An embodiment of the present invention discloses a multi-parameter comprehensive monitoring method for wet desulfurization. Referring to Figure 1 , it includes steps S1 - S3:

[0038] S1: Collect multi-parameter data during the wet desulfurization process.

[0039] It should be noted that the wet desulfurization process is a method for removing sulfur dioxide from gases, mainly applied in fields such as industrial waste gas treatment and flue gas desulfurization in coal-fired power plants. The basic principle of this process is to use a liquid absorbent to react with sulfur dioxide in the gas, thereby reducing the concentration of sulfur dioxide in the waste gas. There are multiple parameters in the wet desulfurization process, and monitoring these parameters is crucial for ensuring the efficiency, safety, and environmental friendliness of the wet desulfurization process. By monitoring these parameters, the desulfurization efficiency can be effectively improved, the operation cost can be reduced, and the negative impact on the environment can be minimized.

[0040] In the embodiment of the present invention, the preset sampling time is 2 seconds per time. During the wet desulfurization process, a spectrometer is arranged to collect the sulfur dioxide concentration at the inlet and outlet of the reaction tower, a pH meter is arranged to collect the pH value of the absorbent solution; a flow meter is arranged to collect the flue gas flow rate; an infrared thermometer is arranged to collect the flue gas temperature. The parameters such as the sulfur dioxide concentration at the inlet and outlet of the reaction tower, the pH value of the absorbent solution, the flue gas flow rate, and the flue gas temperature collected at the same sampling time are recorded as a multi-parameter data. The multi-parameter data at each sampling time is used as a multi-parameter data sequence.

[0041] It should be noted that in the embodiment of the present invention, parameters such as the sulfur dioxide concentration at the inlet and outlet of the reaction tower, the pH value of the absorbent solution, the flue gas flow rate, and the flue gas temperature are collected simultaneously.

[0042] S2: Obtain the characteristic intensity of each parameter in the initial multi-parameter data. According to the characteristic intensity of each parameter in the initial multi-parameter data and the similarity degree of each parameter in the initial multi-parameter data with other parameters, obtain the priority degree of each parameter in the initial multi-parameter data. Based on the priority degree of each parameter, construct an initial isolation tree for the initial multi-parameter data.

[0043] It should be noted that the isolation forest algorithm is an algorithm for anomaly detection of multi-parameter data. Its working principle is based on the idea of "isolation". By selecting one parameter at each layer of the isolation tree to achieve the segmentation of data samples, abnormal data is segmented at the upper layer of the isolation tree, widening the difference between abnormal data and normal data. If any parameter is randomly selected for segmentation at different layers of the isolation tree, it will cause a parameter with poor discrimination to be placed at the upper layer of the isolation tree, making the distance between abnormal data and normal data in the isolation tree relatively close, resulting in inaccurate calculation of the anomaly score. Therefore, it is necessary to calculate the priority of each parameter in the initial multi-parameter data according to the feature strength of each parameter in the initial multi-parameter data, and construct the isolation tree based on the priority of each parameter, which can place the parameter items with good discrimination above the isolation tree, making the subsequent anomaly results more accurate.

[0044] Step S2 includes steps S20 - S21, specifically as follows:

[0045] S20: Obtain the feature strength of each parameter in the initial multi-parameter data.

[0046] It should be noted that in the wet flue gas desulfurization process, different parameters in the multi-parameter data have different change ranges, and different change ranges will affect the segmentation effect of this parameter in the isolation tree. Therefore, it is first necessary to analyze the change range of each parameter in the multi-parameter data, calculate the feature strength of each parameter in the multi-parameter data, and then obtain the priority of each parameter in the multi-parameter data to construct the isolation tree, which can place the features with good discrimination above the isolation tree for subsequent anomaly monitoring;

[0047] It should be further noted that since in the initial stage of the reaction, the initial multi-parameter data provides the reference state of each parameter in the wet flue gas desulfurization process, and these parameters reflect the typical performance of the system during normal operation. Therefore, in the present invention, it is first necessary to obtain the initial multi-parameter data for constructing the isolation tree.

[0048] In the embodiment of the present invention, a preset number of sampling moments X is set, and the multi-parameter data of the first X sampling moments of the multi-parameter data sequence is used as the initial multi-parameter data. In the embodiment of the present invention, the preset sampling moment X = 100. In other embodiments, the implementer can preset the value of the number of sampling moments X according to the specific implementation situation.

[0049] Obtain the feature strength of each parameter in the initial multi-parameter data:

[0050] ;

[0051] where represents the feature strength of the i-th parameter in the initial multi-parameter data; represents the number of the i-th parameter in the initial multi-parameter data; represents the value of the i-th parameter of the j-th in the initial multi-parameter data; represents all the i-th parameters in the initial multi-parameter data; norm() represents the normalization function;

[0052] wherein, represents the difference between each i-th parameter in the initial multi-parameter data and the mean value of all the i-th parameters. The larger its value is, the greater the difference between the i-th parameters in the initial multi-parameter data is, and the greater the feature intensity is; represents the standard deviation of the i-th parameter in the initial multi-parameter data. By dividing by the standard deviation for standardization operation, the influence caused by the different orders of magnitude between different parameters is avoided.

[0053] S21: Obtain the priority of each parameter in the initial multi-parameter data according to the feature intensity of each parameter in the initial multi-parameter data and the similarity degree of each parameter in the initial multi-parameter data with other parameters, and construct an initial isolation tree for the initial multi-parameter data based on the priority of each parameter.

[0054] It should be noted that the greater the feature intensity of a parameter is, the greater the difference between the parameters of this item is, which means that this parameter can effectively distinguish data. Placing it on the upper layer of the isolation tree can split the data earlier, so that abnormal data can be isolated faster. At the same time, due to the mutual influence between some parameters, the similarity degree between each parameter also needs to be considered. If the similarity degree of any parameter with other parameters is relatively low, it means that this parameter has independent information and is not affected by other parameters. Placing this parameter on the upper layer of the isolation tree can help better identify abnormal data. Therefore, the present invention obtains the priority of each parameter in the initial multi-parameter data according to the feature intensity of each parameter in the initial multi-parameter data and the similarity degree of each parameter in the initial multi-parameter data with other parameters.

[0055] In the embodiment of the present invention, obtain the absolute value of the Pearson correlation coefficient between the i-th parameter in the initial multi-parameter data and each parameter except the i-th parameter, and denote the mean value of the absolute values of the Pearson correlation coefficients between the i-th parameter in the initial multi-parameter data and all parameters except the i-th parameter as the similarity degree of the i-th parameter in the initial multi-parameter data with other parameters.

[0056] Obtain the priority of each parameter in the initial multi-parameter data:

[0057] ;

[0058] In the formula, Represents the feature intensity of the i-th parameter in the initial multi-parameter data; Represents the priority of the i-th parameter in the initial multi-parameter data; Represents the similarity degree of the i-th parameter in the initial multi-parameter data with other parameters; among them, the greater the feature intensity of the i-th parameter and the smaller the similarity degree with other parameters, the i-th parameter is used as the upper layer dimension of the isolation tree, that is, the higher the priority of the feature intensity of this parameter; on the contrary, the smaller the feature intensity of the i-th parameter and the greater the similarity degree with other parameters, it is used as the lower layer dimension of the isolation tree, that is, the lower the priority of this parameter.

[0059] Construct the initial isolation tree from top to bottom for the initial multi-parameter data according to the order of the priority of each parameter in the initial multi-parameter data from large to small. It should be noted that the parameter with the largest priority is at the highest layer of the initial isolation tree. Constructing an isolation tree is a prior art, and in the embodiments of the present invention, it will not be elaborated too much here.

[0060] S3: Add the real-time data into the initial isolation tree, calculate the degree of structural change of the initial isolation tree, and adaptively update the initial isolation tree according to the degree of structural change to obtain the final isolation tree; obtain the anomaly score of each multi-parameter data according to the final isolation tree, and give an early warning according to the anomaly score result.

[0061] Step S3 includes steps S30 - S31, specifically as follows:

[0062] S30: Add the real-time data into the initial isolation tree, calculate the degree of structural change of the initial isolation tree, and adaptively update the initial isolation tree according to the degree of structural change to obtain the final isolation tree.

[0063] It should be noted that in the wet flue gas desulfurization process, if the feed of the absorption tower changes (such as the concentration of the absorbent, the flue gas temperature, and the flow rate change), it will cause the distribution of the initial multi-parameter data not to represent the current process state, resulting in inaccurate abnormal results obtained by using the initial isolation tree constructed from the initial multi-parameter data. Therefore, after adding the real-time data into the initial isolation tree, it is necessary to analyze the degree of structural change of the initial isolation tree, so as to adaptively judge whether it is necessary to update the initial isolation tree according to the degree of structural change, and then the abnormal detection results after adding the real-time data into the initial isolation tree can be accurate.

[0064] In an embodiment of the present invention, a preset number M of iteratively updated data is set. The data corresponding to the X-th sampling moment in the multi-parameter data sequence is recorded as the first node in the multi-parameter data sequence. M data after the first node in the multi-parameter data sequence are added to the initial isolation tree to obtain the first updated isolation tree. The ratio of the data of any node in each layer of the first updated isolation tree to the total number of data in that layer is obtained, and the proportion of the number of data of each node in each layer of the first updated isolation tree is obtained. Similarly, the proportion of the number of data of each node in each layer of the initial isolation tree is obtained; the proportion of the number of data is shown in Figure 2 . It should be noted that X is the preset number of sampling moments.

[0065] According to the comparison between the first updated isolation tree and the initial isolation tree, the degree of structural change of the first updated isolation tree is obtained:

[0066] ;

[0067] In the formula, represents the degree of structural change of the first updated isolation tree; represents the number of layers of the first updated isolation tree; represents the proportion of the number of data of the a-th node in the m-th layer of the first updated isolation tree; represents the proportion of the number of data of the a-th node in the m-th layer of the initial isolation tree; represents the layer number sequence of the first updated isolation tree; represents the number of nodes in the m-th layer of the first updated isolation tree; exp() represents the exponential function with the natural constant as the base; || represents the absolute value symbol; represents the corresponding node difference between the initial isolation tree and the first updated isolation tree in all layers. The larger its value, the greater the structural change of the first updated isolation tree; represents the weight of the corresponding node difference between the initial isolation tree and the first updated isolation tree in the m-th layer. The larger its value, the closer the m-th layer is to the root node. If the nodes in the m-th layer change, the greater the impact on the lower layer of the first updated isolation tree, and thus the greater the degree of structural change of the first updated isolation tree; The larger the value of, the greater the degree of structural change of the first updated isolation tree. It is necessary to reconstruct the isolation tree according to the data in the first updated isolation tree.

[0068] The preset change threshold T1 = 0.5. In other embodiments, the implementer can preset the value of the change threshold according to the specific implementation situation. If the degree of structural change of the first updated isolation tree is greater than or equal to the change threshold T1, according to the construction method of the initial isolation tree, reconstruct the isolation tree for all the data in the first updated isolation tree to obtain the first final isolation tree; if the degree of structural change of the first updated isolation tree is less than the change threshold T1, record the first updated isolation tree as the first final isolation tree;

[0069] Record the Mth data after the first node in the multi-parameter data sequence as the second node, and then add the M data after the second node in the multi-parameter data sequence to the first final isolation tree to obtain the second updated isolation tree; according to the first final isolation tree and the second updated isolation tree, obtain the degree of structural change of the second updated isolation tree to determine whether the second updated isolation tree needs to be reconstructed;

[0070] And so on, until there is no data in the multi-parameter data sequence, then stop. At this time, record the pth final isolation tree as the final isolation tree to complete the adaptive update of the isolation tree.

[0071] S31: Obtain the anomaly score of each multi-parameter data according to the final isolation tree, and issue a warning according to the anomaly score result.

[0072] In the embodiment of the present invention, according to the final isolation tree, obtain the anomaly score of each multi-parameter data. If the mean value of the anomaly scores of all multi-parameter data is greater than the preset anomaly threshold T2, the system issues a warning at this time.

[0073] The preset anomaly threshold T2 = 0.8. In other embodiments, the implementer can preset the value of the anomaly threshold T2 according to the specific implementation situation.

[0074] It should be noted that obtaining the anomaly score of each multi-parameter data according to the final isolation tree is a prior art, and in the embodiment of the present invention, it will not be elaborated too much here.

[0075] The embodiment of the present invention also discloses a multi-parameter comprehensive monitoring system for wet flue gas desulfurization, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a multi-parameter comprehensive monitoring method for wet flue gas desulfurization according to the present invention is implemented.

[0076] The above system also includes other components well-known to those skilled in the art such as a communication bus and a communication interface, and their settings and functions are known in the art, so they will not be elaborated here.

[0077] In the present invention, the foregoing memory may be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as, a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid memory cube, etc., or any other medium that can be used to store the required information and can be accessed by an application program, a module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0078] Although this specification has shown and described multiple embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many changes, alterations, and alternative forms will occur to those skilled in the art without departing from the spirit and scope of the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in practicing the present invention.

[0079] The above are all preferred embodiments of the present invention, and the protection scope of the present invention is not limited thereby. Therefore, all equivalent changes made according to the structure, shape, and principle of the present invention shall be covered within the protection scope of the present invention.

Claims

1. A multi-parameter comprehensive monitoring method for wet flue gas desulfurization, characterized in that: Includes steps: Collect initial multi-parameter data; obtain the priority of each parameter in the initial multi-parameter data , Represents the characteristic intensity of the i-th parameter in the initial multi-parameter data; Represents the priority of the i-th parameter in the initial multi-parameter data; represents the similarity between the i-th parameter and other parameters in the initial multi-parameter data; norm() represents a normalization function, which constructs an initial isolation tree for the initial multi-parameter data based on the priority of each parameter; Add the real-time data to the initial isolation tree to obtain the first updated isolation tree; Get the extent of structural changes of the first updated isolated tree ; Represents the level of the first update of the isolated tree; Represents the percentage of data in the first update of the a-th node in the m-th layer of the isolated tree; Represents the number of nodes in the mth layer of the isolated tree updated for the first time; Represents the percentage of data in the a-th node of the m-th layer in the initial isolated tree; Represents the level number of the first update of the isolated tree; exp() represents an exponential function with a natural constant as the base; || represents the absolute value symbol; Adaptively update the initial isolated tree based on the degree of structural change of the first updated isolated tree to obtain the final isolated tree; Anomaly monitoring based on the final isolated tree; The acquisition of the characteristic strength of the i-th parameter in the initial multi-parameter data includes: ; Represents the number of the i-th parameter in the initial multi-parameter data; Represents the value of the i-th parameter of the j-th item in the initial multi-parameter data; Represents all i-th parameters in the initial multi-parameter data.

2. A multi-parameter comprehensive monitoring method for wet flue gas desulfurization according to claim 1, characterized in that: The similarity between the i-th parameter in the initial multi-parameter data and other parameters includes: The absolute value of the Pearson correlation coefficient between the i-th parameter in the initial multi-parameter data and each parameter except the i-th parameter is obtained, and the average of the absolute values ​​of the Pearson correlation coefficients between the i-th parameter in the initial multi-parameter data and all parameters except the i-th parameter is recorded as the similarity between the i-th parameter in the initial multi-parameter data and other parameters.

3. A multi-parameter comprehensive monitoring method for wet flue gas desulfurization according to claim 1, characterized in that: The constructing of an initial isolation tree for the initial multi-parameter data based on the priority of each parameter includes: According to the order of priority of each parameter in the initial multi-parameter data from large to small, an initial isolation tree is constructed from top to bottom for the initial multi-parameter data.

4. A multi-parameter comprehensive monitoring method for wet flue gas desulfurization according to claim 1, characterized in that: The step of adding the real-time data to the initial isolation tree to obtain the first updated isolation tree includes: The number of iterative update data is preset to M, the data corresponding to the Xth sampling moment in the multi-parameter data sequence is recorded as the first node in the multi-parameter data sequence, and the M data after the first node are added to the initial isolated tree to obtain the first updated isolated tree.

5. The multi-parameter comprehensive monitoring method for wet flue gas desulfurization according to claim 1, characterized in that: The obtaining of the final isolated tree comprises: The number of iterative update data is preset M, and the change threshold T1 is preset. If the degree of structural change of the first updated isolated tree is greater than or equal to T1, the isolated tree is rebuilt for all data in the first updated isolated tree according to the construction method of the initial isolated tree to obtain the first final isolated tree; otherwise, the first updated isolated tree is recorded as the first final isolated tree; the Mth data after the first node in the multi-parameter data sequence is recorded as the second node, and the M data after the second node are added to the first final isolated tree to obtain the second updated isolated tree; the degree of structural change of the second updated isolated tree is obtained to determine whether the second updated isolated tree needs to be rebuilt; and so on, until there is no data in the multi-parameter data sequence, and the pth final isolated tree is recorded as the final isolated tree.

6. A multi-parameter comprehensive monitoring method for wet flue gas desulfurization according to claim 1, characterized in that: The abnormality monitoring according to the final isolated tree includes: The abnormal threshold T2 is preset, and the abnormal score of each multi-parameter data is obtained according to the final isolation tree. If the mean of the abnormal scores of all multi-parameter data is greater than the preset abnormal threshold T2, the system issues an early warning.

7. A multi-parameter comprehensive monitoring system for wet flue gas desulfurization, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a multi-parameter comprehensive monitoring method for wet desulfurization according to any one of claims 1 to 6 is implemented.

Images