CN116934558A - Automatic patrol monitoring method and system for unmanned aerial vehicle - Google Patents
Automatic patrol monitoring method and system for unmanned aerial vehicle Download PDFInfo
- Publication number
- CN116934558A CN116934558A CN202311200125.4A CN202311200125A CN116934558A CN 116934558 A CN116934558 A CN 116934558A CN 202311200125 A CN202311200125 A CN 202311200125A CN 116934558 A CN116934558 A CN 116934558A
- Authority
- CN
- China
- Prior art keywords
- water quality
- river reach
- data
- predicted
- historical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000012544 monitoring process Methods 0.000 title claims abstract description 36
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 172
- 230000004927 fusion Effects 0.000 claims abstract description 36
- 238000007689 inspection Methods 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims abstract description 20
- 230000007246 mechanism Effects 0.000 claims description 35
- 238000013528 artificial neural network Methods 0.000 claims description 34
- 238000012549 training Methods 0.000 claims description 34
- 239000013598 vector Substances 0.000 claims description 27
- 238000013507 mapping Methods 0.000 claims description 16
- 238000011156 evaluation Methods 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 15
- 238000012512 characterization method Methods 0.000 claims description 13
- 238000007781 pre-processing Methods 0.000 claims description 10
- 230000004931 aggregating effect Effects 0.000 claims description 8
- 238000010586 diagram Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 6
- 230000009286 beneficial effect Effects 0.000 abstract description 5
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 6
- 229910052760 oxygen Inorganic materials 0.000 description 6
- 239000001301 oxygen Substances 0.000 description 6
- 230000008901 benefit Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 239000013505 freshwater Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000009428 plumbing Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012876 topography Methods 0.000 description 2
- -1 turbidity Substances 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The application provides an automatic patrol monitoring method and system for an unmanned aerial vehicle, wherein the method comprises the following steps: constructing a knowledge graph; collecting data; obtaining a historical water quality characteristic value of the fusion knowledge; inputting the data into a model for prediction; extracting features in the predicted river reach image, and fusing the features with predicted water quality feature values; the map of the inspection area is divided according to the maximum value of the probability of the parameter category by processing and analyzing the fused data, so that a key inspection area and a non-key inspection area are obtained; the unmanned aerial vehicle selects the inspection mode and frequency according to the important inspection area and the non-important inspection area. The method provided by the application combines deeper time dimension characteristics and image data characteristics, has higher accuracy and reliability of the prediction result, can capture the influence of external factors on water quality, improves the prediction capability of water quality disasters in rivers and lakes, and is beneficial to improving the inspection efficiency and monitoring effect.
Description
Technical Field
The application relates to the technical field of data monitoring, in particular to an automatic patrol monitoring method and system for an unmanned aerial vehicle.
Background
With the continuous development of unmanned aerial vehicle technology, unmanned aerial vehicles are increasingly widely used in various fields. Especially in the field of river and lake patrol and monitoring, unmanned aerial vehicle can replace the manpower to carry out high-efficient, real-time operation to real-time supervision river and lake water quality disaster condition.
Prediction of water quality disaster conditions in rivers and lakes is often regarded as a time series prediction problem, and previous attention mechanisms mainly focus on the influence of time series at different moments on prediction, but are influenced by various internal and external factors, so that the time dynamics of the water quality disaster time series are complex and changeable, the accuracy and reliability of the water quality disaster prediction result are poor, and the influence of external factors is less considered in prediction.
Therefore, the application aims to provide an automatic patrol monitoring method and system for an unmanned aerial vehicle, so as to solve the problems in the prior art.
Disclosure of Invention
The application aims to provide an unmanned aerial vehicle automatic patrol monitoring method and system thereof, so as to strengthen the accuracy and reliability of a water quality disaster prediction result, fully consider the influence of external factors during prediction and improve the prediction capability of water quality disasters in rivers and lakes.
In a first aspect, an embodiment of the present application provides an automatic unmanned aerial vehicle patrol monitoring method, including the following steps:
constructing a knowledge graph of water quality prediction;
collecting historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach, and determining entities and relations;
encoding the entity and the relation, and simultaneously carrying out knowledge representation;
fusing the knowledge obtained in the steps with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge;
respectively inputting the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period;
the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample;
the second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples;
extracting features in the predicted river reach image, and fusing the features with the predicted water quality feature value to obtain a comprehensive predicted data feature vector, wherein a specific fusion expression is as follows:
Y=Concat(g1*S,g2*T)；
Wherein g1 is used for controlling the predicted water quality characteristic weight, g2 is used for controlling the predicted river reach image characteristic weight, the sum of the predicted water quality characteristic weight and the predicted river reach image characteristic weight is 1, Y is a comprehensive predicted data characteristic vector, S is a predicted water quality characteristic, and T is a predicted river reach image characteristic;
and the water quality of the river reach is evaluated by processing and analyzing the fused data.
The method for acquiring the historical water quality characteristic data and the historical river reach image data of the river reach patrol area, determining interest points owned by the river reach and determining the entity and the relationship comprises the following steps:
acquiring historical water quality characteristic data of a river reach patrol area;
obtaining interest points and the number thereof owned by a river reach, and dividing the number of each interest point into three types of less, medium and more according to 33% of the number of the interest points owned by the river reach and 66% of the number of the interest points owned by the river reach;
determining the number of entities including river reach and interest points according to the existing data;
and determining a relation according to the existing data, wherein the relation comprises a river reach adjacency relation and a point of interest of the river reach.
The method for encoding the entities and the relationships and simultaneously carrying out knowledge representation comprises the following steps:
coding the quantity of each river reach and each point of interest according to the determined entity;
Coding the adjacent relationship of the river reach and the interest points of the river reach according to the determined relationship;
constructing a knowledge graph triplet according to the coded entities and the relation;
knowledge representation is carried out by using a TransE method, and vector representation X of characteristics of all the river reach containing interest points is obtained E 。
The knowledge obtained in the steps is fused with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge, and the method comprises the following steps:
the vector representation X of each river reach observed at the moment t and containing the interesting point features E And historical water quality characteristic X t The method is characterized in that the method comprises the steps of inputting into a main space-time diagram convolution network, outputting a historical water quality characteristic value X after fusion of a moment t and external knowledge r The expression is:
X r =Relu(X E X t W r +b r )；
wherein Relu is an activation function for deep learning, W r Is a linear transformation coefficient, b r Is the offset.
The method pre-trains the first predictive model as follows:
acquiring a historical water quality characteristic value;
after preprocessing the historical water quality characteristic value sample, converting the historical water quality characteristic data sample into water quality characteristic value supervised data;
and inputting the water quality characteristic value supervised data into the neural network based on the first multi-directional attention mechanism for training, and obtaining the first prediction model after reaching a preset training cut-off condition.
The first neural network based on the multidirectional attention mechanism comprises a first linear layer, a first normalization layer, a first time representation fusion layer and a first output layer which are sequentially connected;
a first linear layer for mapping water quality characteristic value supervised data to a multi-directional attention importance score representation;
the first normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the first time representation fusion layer is used for aggregating the water quality characteristic value supervised data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a first output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
The method pre-trains the second predictive model as follows:
acquiring historical river reach image data;
preprocessing a historical river reach image data sample, and converting the historical river reach image data sample into river reach image data supervised data;
inputting the supervised data of the river reach image data into the neural network based on the second multidirectional attention mechanism for training, and obtaining the second prediction model after reaching a preset training cut-off condition.
The second neural network based on the multidirectional attention mechanism comprises a second linear layer, a second normalization layer, a second time representation fusion layer and a second output layer which are sequentially connected;
a second linear layer for mapping the supervised data of the river reach image data to a multi-directional attention importance score representation;
the second normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the second time representation fusion layer is used for aggregating the supervised data of the river reach image data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a second output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
The extraction of the characteristics in the predicted river reach image comprises the steps of using multispectral images or hyperspectral images and identifying the water quality type according to the reflection spectrum characteristics of different water quality types.
The method realizes the assessment of the water quality of the river reach by processing and analyzing the fused data, and comprises the following steps:
processing the feature vector Y of the comprehensive prediction data through a linear layer and a softMax layer to obtain probability weights of each comprehensive prediction data on each parameter class;
Outputting the maximum value of the probability of the parameter category of each comprehensive prediction data;
dividing a map of the inspection area according to the maximum value of the parameter class probability to obtain an important inspection area and a non-important inspection area;
the unmanned aerial vehicle selects the inspection mode and frequency according to the important inspection area and the non-important inspection area.
In a second aspect, an embodiment of the present application provides an automatic unmanned aerial vehicle patrol monitoring system, including:
the construction module is used for constructing a knowledge graph of water quality prediction;
the acquisition module is used for acquiring historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach and determining entities and relations;
the coding module is used for coding the entity and the relation and simultaneously carrying out knowledge representation;
the first fusion module fuses the obtained knowledge with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge;
the processing module is used for respectively inputting the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period;
the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample;
The second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples;
the second fusion module is used for extracting the characteristics in the predicted river reach image, and fusing the characteristics with the predicted water quality characteristic value to obtain a comprehensive predicted data characteristic vector, wherein the specific fusion expression is as follows:
Y=Concat(g1*S,g2*T)；
wherein g1 is used for controlling the predicted water quality characteristic weight, g2 is used for controlling the predicted river reach image characteristic weight, the sum of the predicted water quality characteristic weight and the predicted river reach image characteristic weight is 1, Y is a comprehensive predicted data characteristic vector, S is a predicted water quality characteristic, and T is a predicted river reach image characteristic;
and the evaluation module is used for realizing the evaluation of the water quality of the river reach by processing and analyzing the fused data.
Compared with the prior art, the unmanned aerial vehicle automatic patrol monitoring method provided by the application has the advantages that a knowledge graph of water quality prediction is constructed; collecting historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach, and determining entities and relations; encoding the entity and the relation, and simultaneously carrying out knowledge representation; fusing the knowledge obtained in the steps with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge; respectively inputting the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period; the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample; the second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples; extracting features in the predicted river reach image, and fusing the features with predicted water quality feature values to obtain comprehensive predicted data feature vectors; and the water quality of the river reach is evaluated by processing and analyzing the fused data. Compared with the traditional prediction method, the method provided by the application combines the deeper time dimension characteristics and the image data characteristics, has higher accuracy and reliability of the prediction result, can capture the influence of external factors on water quality, improves the prediction capability of water quality disasters in rivers and lakes, and is beneficial to improving the inspection efficiency and monitoring effect.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the application. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:
fig. 1 shows a flowchart of an automatic unmanned aerial vehicle patrol monitoring method provided by the application;
fig. 2 shows a schematic diagram of an automatic unmanned aerial vehicle patrol monitoring system provided by the application;
fig. 3 shows a schematic diagram of an electronic device provided by the application.
Description of the embodiments
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
It is noted that unless otherwise indicated, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this application belongs.
In addition, the terms "first" and "second" etc. are used to distinguish different objects and are not used to describe a particular order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus.
Referring to fig. 1, fig. 1 is a flowchart of an automatic unmanned aerial vehicle inspection and monitoring method according to an embodiment of the present application, including the following steps S101 to S102:
s101, constructing a knowledge graph of water quality prediction;
s102, acquiring historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach, and determining entities and relations; specifically, the sensor and the collector can be carried on the unmanned aerial vehicle, and various types of sensors and collectors can be carried on the unmanned aerial vehicle aiming at the environmental monitoring of the freshwater lake, including a high-definition camera, a multispectral camera, an infrared measurement device, a water quality sensor and the like, so as to realize comprehensive monitoring;
The historical water quality characteristic data can be index data such as pH value, dissolved oxygen, turbidity, chemical oxygen demand and the like, and the historical river reach image data can be image data such as aquatic organisms, river topography, hydrologic characteristics (color textures) and the like.
In order to make the final predicted data more accurate, the step of preprocessing the historical river reach image data is further included after the historical river reach image data of the river reach patrol area is collected: the method comprises the operations of historical river reach image enhancement, denoising, filtering and the like so as to ensure the accuracy and comparability of data.
In this embodiment, the point of interest POI specifically may be: industry, agriculture, life, plumbing, temperature, etc., and divides the number of points of interest per POI into three categories of less, medium, more (classification can be freely set), expressed as less/medium/more + POI point names, with intermediate underlined connections, such as less _ industry, the number of points of interest representing the name of the river reach is within 33% of the quantile of industry (including 33%), the middle industry represents the number of points of interest representing the name of the river reach is between 33% of the quantile of industry and 66% of the quantile of industry (excluding 33% of the number of points of interest and including 66%), and the multiple industry represents the number of points of interest representing the name of the river reach is above 66% of the quantile of industry (excluding 66%).
S103, encoding the entity and the relation and simultaneously carrying out knowledge representation, wherein the method specifically comprises the following steps:
s1031, coding the quantity of each river reach and each point of interest according to the determined entity;
s1032, coding the adjacent relationship of the river reach and the interest points of the river reach according to the determined relationship;
s1033, constructing a knowledge graph triplet according to the coded entities and the relations; for example, triples that can be constructed are: para 1, have_industry_point of interest POI, poly_industry, expressed in code: [1, 11,2].
S1034, carrying out knowledge representation by using a TransE method to obtain vector representation X of each river reach containing interest point characteristics E . Wherein, the TransE is a knowledge graph embedding method.
S104, fusing the knowledge obtained in the step with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge;
in some embodiments, the fusing the knowledge obtained in the above steps with the historical water quality characteristic data to obtain the historical water quality characteristic value of the fused knowledge includes the following steps:
the vector representation X of each river reach observed at the moment t and containing the interesting point features E And historical water quality characteristic X t The method is characterized in that the method comprises the steps of inputting into a main space-time diagram convolution network, outputting a historical water quality characteristic value X after fusion of a moment t and external knowledge r The expression is:
X r =Relu(X E X t W r +b r )；
wherein Relu is an activation function for deep learning, W r Is a linear transformation coefficient, b r Is the offset;
the obtained historical water quality characteristic value of the fusion knowledge is more accurate and reliable.
S105, respectively inputting the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period;
the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample;
the following describes how to pre-train to obtain the first prediction model, specifically, the following manner may be adopted:
s1051, acquiring a historical water quality characteristic value;
s1052, after preprocessing the historical water quality characteristic value sample, converting the historical water quality characteristic data sample into water quality characteristic value supervised data;
s1053, inputting the water quality characteristic value supervised data into the neural network based on the first multidirectional attention mechanism for training, and obtaining the first prediction model after reaching a preset training cut-off condition.
Specifically, the first neural network based on the multi-directional attention mechanism comprises a first linear layer, a first normalization layer, a first time characterization fusion layer and a first output layer which are sequentially connected;
a first linear layer for mapping water quality characteristic value supervised data to a multi-directional attention importance score representation;
the first normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the first time representation fusion layer is used for aggregating the water quality characteristic value supervised data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a first output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
The second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples;
the following describes how the second predictive model may be trained in advance, in particular in the following manner:
s1054, acquiring historical river reach image data;
s1055, preprocessing the historical river reach image data sample, and converting the historical river reach image data sample into river reach image data supervised data;
S1056, inputting the supervised data of the river reach image data into the neural network based on the second multidirectional attention mechanism for training, and obtaining the second prediction model after reaching a preset training cut-off condition.
Specifically, the second neural network based on the multi-directional attention mechanism comprises a second linear layer, a second normalization layer, a second time characterization fusion layer and a second output layer which are sequentially connected;
a second linear layer for mapping the supervised data of the river reach image data to a multi-directional attention importance score representation;
the second normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the second time representation fusion layer is used for aggregating the supervised data of the river reach image data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a second output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
S106, extracting features in the predicted river reach image, and fusing the features with the predicted water quality feature value to obtain a comprehensive predicted data feature vector, wherein a specific fusion expression is as follows:
Y=Concat(g1*S,g2*T)；
Wherein g1 is used for controlling the predicted water quality characteristic weight, g2 is used for controlling the predicted river reach image characteristic weight, the sum of the predicted water quality characteristic weight and the predicted river reach image characteristic weight is 1, Y is a comprehensive predicted data characteristic vector, S is a predicted water quality characteristic, T is a predicted river reach image characteristic, and g1 and g2 can be set according to actual conditions and only the conditions are met;
in S106, the extracting the features in the predicted river reach image includes identifying the water quality type according to the reflection spectrum features of different water quality types by using the multispectral image or the hyperspectral image. The above features include geometric features including the length, width, curvature, flow rate, etc. of the river, and physical features including dissolved oxygen, turbidity, pH, conductivity, etc. of the river.
S107, the fused data are processed and analyzed to realize the assessment of the water quality of the river reach.
There are various methods for evaluation, such as combining expert systems, support Vector Machines (SVMs), etc., to perform comprehensive analysis and evaluation. The evaluation method of this embodiment may specifically be: and processing the characteristic vector Y of the comprehensive predicted data through a linear layer and a softMax layer to obtain the probability weight of each comprehensive predicted data on each parameter class, and outputting the maximum value of the probability of the parameter class of each comprehensive predicted data to realize the evaluation of water quality. For example, the water quality of the river reach can be evaluated by predicting the pH value of the river reach to 7 with a probability of eighty percent and predicting the turbidity of the river reach to 200 with a probability of seventy percent.
In the application, the map of the inspection area is divided according to the maximum value of the probability of the parameter category to obtain a key inspection area and a non-key inspection area; the unmanned aerial vehicle selects the mode and the frequency of patrolling according to the important area of patrolling and non-important area of patrolling and examining, for example unmanned aerial vehicle flight's height, angle, time and the frequency of patrolling and examining, the important area of patrolling and examining can increase the frequency of patrolling and examining, and the non-important area of patrolling and examining can reduce the frequency of patrolling and examining to improve efficiency and monitoring effect of patrolling and examining.
Compared with the prior art, the unmanned aerial vehicle automatic patrol monitoring method provided by the application has the advantages that a knowledge graph of water quality prediction is constructed; collecting historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach, and determining entities and relations; encoding the entity and the relation, and simultaneously carrying out knowledge representation; fusing the knowledge obtained in the steps with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge; respectively inputting the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period; the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample; the second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples; extracting features in the predicted river reach image, and fusing the features with predicted water quality feature values to obtain comprehensive predicted data feature vectors; and the water quality of the river reach is evaluated by processing and analyzing the fused data. Compared with the traditional prediction method, the method provided by the application combines the deeper time dimension characteristics and the image data characteristics, has higher accuracy and reliability of the prediction result, can capture the influence of external factors on water quality, improves the prediction capability of water quality disasters in rivers and lakes, and is beneficial to improving the inspection efficiency and monitoring effect.
In the above embodiment, an automatic unmanned aerial vehicle patrol monitoring method is provided, and correspondingly, the application further provides an automatic unmanned aerial vehicle patrol monitoring system. The unmanned aerial vehicle automatic patrol monitoring system provided by the embodiment of the application can implement the unmanned aerial vehicle automatic patrol monitoring method, and the monitoring system can be realized by software, hardware or a combination of software and hardware. For example, the monitoring system may include integrated or separate functional modules or units to perform the corresponding steps in the methods described above, including:
the construction module 101 is used for constructing a knowledge graph of water quality prediction;
the acquisition module 102 is used for acquiring historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach and determining entities and relations; specifically, the sensor and the collector can be carried on the unmanned aerial vehicle, and various types of sensors and collectors can be carried on the unmanned aerial vehicle aiming at the environmental monitoring of the freshwater lake, including a high-definition camera, a multispectral camera, an infrared measurement device, a water quality sensor and the like, so as to realize comprehensive monitoring;
the historical water quality characteristic data can be index data such as pH value, dissolved oxygen, turbidity, chemical oxygen demand and the like, and the historical river reach image data can be image data such as aquatic organisms, river topography, hydrologic characteristics (color textures) and the like.
In order to make the final predicted data more accurate, the method further comprises a historical river reach image data preprocessing module after the historical river reach image data of the river reach patrol area is collected: the method comprises the operations of historical river reach image enhancement, denoising, filtering and the like so as to ensure the accuracy and comparability of data.
In this embodiment, the point of interest POI specifically may be: industry, agriculture, life, plumbing, temperature, etc., and divides the number of points of interest per POI into three categories of less, medium, more (classification can be freely set), expressed as less/medium/more + POI point names, with intermediate underlined connections, such as less _ industry, the number of points of interest representing the name of the river reach is within 33% of the quantile of industry (including 33%), the middle industry represents the number of points of interest representing the name of the river reach is between 33% of the quantile of industry and 66% of the quantile of industry (excluding 33% of the number of points of interest and including 66%), and the multiple industry represents the number of points of interest representing the name of the river reach is above 66% of the quantile of industry (excluding 66%).
An encoding module 103, configured to encode the entity and the relationship, and perform knowledge representation at the same time; it can be used for:
Coding the quantity of each river reach and each point of interest according to the determined entity;
coding the adjacent relationship of the river reach and the interest points of the river reach according to the determined relationship;
constructing a knowledge graph triplet according to the coded entities and the relation; for example, triples that can be constructed are: para 1, have_industry_point of interest POI, poly_industry, expressed in code: [1, 11,2].
Knowledge representation is carried out by using a TransE method, and vector representation X of characteristics of all the river reach containing interest points is obtained E . Wherein, the TransE is a knowledge graph embedding method.
The first fusion module 104 fuses the obtained knowledge with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge;
in some embodiments, the fusing the obtained knowledge with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge specifically includes:
the vector representation X of each river reach observed at the moment t and containing the interesting point features E And historical water quality characteristic X t The method is characterized in that the method comprises the steps of inputting into a main space-time diagram convolution network, outputting a historical water quality characteristic value X after fusion of a moment t and external knowledge r The expression is:
X r =Relu(X E X t W r +b r )；
Wherein Relu is an activation function for deep learning, W r Is a linear transformation coefficient, b r Is the offset;
the obtained historical water quality characteristic value of the fusion knowledge is more accurate and reliable.
The processing module 105 respectively inputs the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period;
the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample;
the second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples;
the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample;
the following describes how to pre-train to obtain the first prediction model, specifically, the following manner may be adopted:
acquiring a historical water quality characteristic value;
after preprocessing the historical water quality characteristic value sample, converting the historical water quality characteristic data sample into water quality characteristic value supervised data;
And inputting the water quality characteristic value supervised data into the neural network based on the first multi-directional attention mechanism for training, and obtaining the first prediction model after reaching a preset training cut-off condition.
Specifically, the first neural network based on the multi-directional attention mechanism comprises a first linear layer, a first normalization layer, a first time characterization fusion layer and a first output layer which are sequentially connected;
a first linear layer for mapping water quality characteristic value supervised data to a multi-directional attention importance score representation;
the first normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the first time representation fusion layer is used for aggregating the water quality characteristic value supervised data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a first output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
The second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples;
the following describes how the second predictive model may be trained in advance, in particular in the following manner:
Acquiring historical river reach image data;
preprocessing a historical river reach image data sample, and converting the historical river reach image data sample into river reach image data supervised data;
inputting the supervised data of the river reach image data into the neural network based on the second multidirectional attention mechanism for training, and obtaining the second prediction model after reaching a preset training cut-off condition.
Specifically, the second neural network based on the multi-directional attention mechanism comprises a second linear layer, a second normalization layer, a second time characterization fusion layer and a second output layer which are sequentially connected;
a second linear layer for mapping the supervised data of the river reach image data to a multi-directional attention importance score representation;
the second normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the second time representation fusion layer is used for aggregating the supervised data of the river reach image data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a second output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
The second fusion module 106 is configured to extract features in the predicted river reach image, and fuse the features with the predicted water quality feature value to obtain a comprehensive predicted data feature vector, where a specific fusion expression is as follows:
Y=Concat(g1*S,g2*T)；
Wherein g1 is used for controlling the predicted water quality characteristic weight, g2 is used for controlling the predicted river reach image characteristic weight, the sum of the predicted water quality characteristic weight and the predicted river reach image characteristic weight is 1, Y is a comprehensive predicted data characteristic vector, S is a predicted water quality characteristic, and T is a predicted river reach image characteristic;
in the second fusion module 106, the extracting the features in the predicted river reach image includes identifying the water quality type according to the reflection spectrum features of different water quality types by using the multispectral image or the hyperspectral image. The above features include geometric features including the length, width, curvature, flow rate, etc. of the river, and physical features including dissolved oxygen, turbidity, pH, conductivity, etc. of the river.
And the evaluation module 107 is used for realizing the evaluation of the water quality of the river reach by processing and analyzing the fused data.
There are various kinds of evaluation modules, such as a combination of expert systems, support Vector Machines (SVMs), etc., for comprehensive analysis and evaluation. The evaluation module 107 of this embodiment may specifically be: and processing the characteristic vector Y of the comprehensive predicted data through a linear layer and a softMax layer to obtain the probability weight of each comprehensive predicted data on each parameter class, and outputting the maximum value of the probability of the parameter class of each comprehensive predicted data to realize the evaluation of water quality. For example, the water quality of the river reach can be evaluated by predicting the pH value of the river reach to 7 with a probability of eighty percent and predicting the turbidity of the river reach to 200 with a probability of seventy percent.
In the application, the map of the inspection area is divided according to the maximum value of the probability of the parameter category to obtain a key inspection area and a non-key inspection area; the unmanned aerial vehicle selects the mode and the frequency of patrolling according to the important area of patrolling and non-important area of patrolling and examining, for example unmanned aerial vehicle flight's height, angle, time and the frequency of patrolling and examining, the important area of patrolling and examining can increase the frequency of patrolling and examining, and the non-important area of patrolling and examining can reduce the frequency of patrolling and examining to improve efficiency and monitoring effect of patrolling and examining.
The monitoring system provided by the embodiment of the application and the monitoring method provided by the embodiment of the application have the same beneficial effects as the method adopted, operated or realized by the monitoring system and the monitoring method provided by the embodiment of the application due to the same inventive concept.
The embodiment of the application also provides an electronic device corresponding to the method provided by the previous embodiment, wherein the electronic device can be an electronic device for a client, such as a mobile phone, a notebook computer, a tablet computer, a desktop computer and the like, so as to execute the prediction method.
Referring to fig. 3, a schematic diagram of an electronic device according to some embodiments of the present application is shown. As shown in fig. 3, the electronic device 20 includes: a processor 200, a memory 201, a bus 202 and a communication interface 203, the processor 200, the communication interface 203 and the memory 201 being connected by the bus 202; the memory 201 stores a computer program executable on the processor 200, and the processor 200 executes the method according to any of the foregoing embodiments of the present application when the computer program is executed.
The memory 201 may include a high-speed random access memory (RAM: random Access Memory), and may further include a non-volatile memory (non-volatile memory), such as at least one disk memory. The communication connection between the system network element and at least one other network element is implemented via at least one communication interface 203 (which may be wired or wireless), the internet, a wide area network, a local network, a metropolitan area network, etc. may be used.
Bus 202 may be an ISA bus, a PCI bus, an EISA bus, or the like. The buses may be classified as address buses, data buses, control buses, etc. The memory 201 is configured to store a program, and the processor 200 executes the program after receiving an execution instruction, and the method disclosed in any of the foregoing embodiments of the present application may be applied to the processor 200 or implemented by the processor 200.
The processor 200 may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in the processor 200 or by instructions in the form of software. The processor 200 may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP for short), etc.; but may also be a Digital Signal Processor (DSP), application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. The disclosed methods, steps, and logic blocks in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in the memory 201, and the processor 200 reads the information in the memory 201, and in combination with its hardware, performs the steps of the above method.
The electronic equipment provided by the embodiment of the application and the prediction method provided by the embodiment of the application have the same beneficial effects as the method adopted, operated or realized by the electronic equipment based on the same inventive concept.
The present application also provides a computer readable storage medium corresponding to the prediction method provided in the foregoing embodiment, on which a computer program (i.e. a program product) is stored, which when executed by a processor, performs the prediction method provided in any of the foregoing embodiments.
It should be noted that examples of the computer readable storage medium may also include, but are not limited to, a phase change memory (PRAM), a Static Random Access Memory (SRAM), a Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), a Read Only Memory (ROM), an Electrically Erasable Programmable Read Only Memory (EEPROM), a flash memory, or other optical or magnetic storage medium, which will not be described in detail herein.
The computer-readable storage medium provided by the above-described embodiments of the present application has the same advantageous effects as the method adopted, operated or implemented by the application program stored therein, for the same inventive concept as the prediction method provided by the embodiments of the present application.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the application, and are intended to be included within the scope of the appended claims and description.
Claims (10)
1. The automatic patrol monitoring method for the unmanned aerial vehicle is characterized by comprising the following steps of:
constructing a knowledge graph of water quality prediction;
collecting historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach, and determining entities and relations;
encoding the entity and the relation, and simultaneously carrying out knowledge representation;
fusing the knowledge obtained in the steps with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge;
respectively inputting the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period;
The first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample;
the second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples;
extracting features in the predicted river reach image, and fusing the features with the predicted water quality feature value to obtain a comprehensive predicted data feature vector, wherein a specific fusion expression is as follows:
Y=Concat(g1*S,g2*T)；
wherein g1 is used for controlling the predicted water quality characteristic weight, g2 is used for controlling the predicted river reach image characteristic weight, the sum of the predicted water quality characteristic weight and the predicted river reach image characteristic weight is 1, Y is a comprehensive predicted data characteristic vector, S is a predicted water quality characteristic, and T is a predicted river reach image characteristic;
and the water quality of the river reach is evaluated by processing and analyzing the fused data.
2. The method according to claim 1, wherein the steps of collecting historical water quality characteristic data and historical river reach image data of a river reach patrolling area, determining interest points owned by the river reach, and determining entities and relations include the following steps:
acquiring historical water quality characteristic data of a river reach patrol area;
Obtaining interest points and the number thereof owned by a river reach, and dividing the number of each interest point into three types of less, medium and more according to 33% of the number of the interest points owned by the river reach and 66% of the number of the interest points owned by the river reach;
determining the number of entities including river reach and interest points according to the existing data;
and determining a relation according to the existing data, wherein the relation comprises a river reach adjacency relation and a point of interest of the river reach.
3. The method of claim 2, wherein the encoding of entities and relationships, while simultaneously representing knowledge, comprises the steps of:
coding the quantity of each river reach and each point of interest according to the determined entity;
coding the adjacent relationship of the river reach and the interest points of the river reach according to the determined relationship;
constructing a knowledge graph triplet according to the coded entities and the relation;
knowledge representation is carried out by using a TransE method, and vector representation X of characteristics of all the river reach containing interest points is obtained E 。
4. A method according to claim 3, wherein the step of fusing knowledge obtained in the above step with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge comprises the steps of:
the vector representation X of each river reach observed at the moment t and containing the interesting point features E And historical water quality characteristic X t The method is characterized in that the method comprises the steps of inputting into a main space-time diagram convolution network, outputting a historical water quality characteristic value X after fusion of a moment t and external knowledge r The expression is:
X r =Relu(X E X t W r +b r )；
wherein Relu is an activation function for deep learning, W r Is a linear transformation coefficient, b r Is the offset.
5. The method of claim 1, wherein the first predictive model is pre-trained in the following manner:
acquiring a historical water quality characteristic value;
after preprocessing the historical water quality characteristic value sample, converting the historical water quality characteristic data sample into water quality characteristic value supervised data;
and inputting the water quality characteristic value supervised data into the neural network based on the first multi-directional attention mechanism for training, and obtaining the first prediction model after reaching a preset training cut-off condition.
6. The method of claim 5, wherein the first multi-directional attention mechanism based neural network comprises a first linear layer, a first normalization layer, a first time characterization fusion layer, and a first output layer connected in sequence;
a first linear layer for mapping water quality characteristic value supervised data to a multi-directional attention importance score representation;
The first normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the first time representation fusion layer is used for aggregating the water quality characteristic value supervised data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a first output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
7. The method of claim 1, wherein the second predictive model is pre-trained in the following manner:
acquiring historical river reach image data;
preprocessing a historical river reach image data sample, and converting the historical river reach image data sample into river reach image data supervised data;
inputting the supervised data of the river reach image data into the neural network based on the second multidirectional attention mechanism for training, and obtaining the second prediction model after reaching a preset training cut-off condition.
8. The method of claim 7, wherein the second multi-directional attention mechanism based neural network comprises a second linear layer, a second normalized layer, a second time-characterized fusion layer, and a second output layer connected in sequence;
A second linear layer for mapping the supervised data of the river reach image data to a multi-directional attention importance score representation;
the second normalization layer is used for normalizing the importance degree score representation of the multi-directional attention to generate a multi-directional attention representation;
the second time representation fusion layer is used for aggregating the supervised data of the river reach image data and the multi-directional attention representation to obtain a cascade representation for fusing the multi-directional attention representation;
a second output layer for mapping the cascade representation of the fused multi-directional attention characterizations to predicted values.
9. The method of claim 1, wherein the evaluation of the water quality of the river reach is achieved by processing and analyzing the fused data, comprising:
processing the feature vector Y of the comprehensive prediction data through a linear layer and a softMax layer to obtain probability weights of each comprehensive prediction data on each parameter class;
outputting the maximum value of the probability of the parameter category of each comprehensive prediction data;
dividing a map of the inspection area according to the maximum value of the parameter class probability to obtain an important inspection area and a non-important inspection area;
the unmanned aerial vehicle selects the inspection mode and frequency according to the important inspection area and the non-important inspection area.
10. An unmanned aerial vehicle automatic patrol monitoring system, which is characterized by comprising:
the construction module is used for constructing a knowledge graph of water quality prediction;
the acquisition module is used for acquiring historical water quality characteristic data and historical river reach image data of a river reach patrol area, determining interest points owned by the river reach and determining entities and relations;
the coding module is used for coding the entity and the relation and simultaneously carrying out knowledge representation;
the first fusion module fuses the obtained knowledge with the historical water quality characteristic data to obtain a historical water quality characteristic value of the fused knowledge;
the processing module is used for respectively inputting the historical water quality characteristic value and the historical river reach image data into a first prediction model and a second prediction model which are trained in advance to obtain the predicted water quality characteristic value and the predicted river reach image data in a future set time period;
the first prediction model is obtained by training a first neural network based on a multidirectional attention mechanism through a historical water quality characteristic value sample;
the second prediction model is obtained by training a second neural network based on a multidirectional attention mechanism through historical river reach image data samples;
the second fusion module is used for extracting the characteristics in the predicted river reach image, and fusing the characteristics with the predicted water quality characteristic value to obtain a comprehensive predicted data characteristic vector, wherein the specific fusion expression is as follows:
Y=Concat(g1*S,g2*T)；
Wherein g1 is used for controlling the predicted water quality characteristic weight, g2 is used for controlling the predicted river reach image characteristic weight, the sum of the predicted water quality characteristic weight and the predicted river reach image characteristic weight is 1, Y is a comprehensive predicted data characteristic vector, S is a predicted water quality characteristic, T is a predicted river reach image characteristic, Y is a comprehensive predicted data characteristic vector, S is a predicted water quality characteristic, and T is a predicted river reach image characteristic;
and the evaluation module is used for realizing the evaluation of the water quality of the river reach by processing and analyzing the fused data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311200125.4A CN116934558B (en) | 2023-09-18 | 2023-09-18 | Automatic patrol monitoring method and system for unmanned aerial vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311200125.4A CN116934558B (en) | 2023-09-18 | 2023-09-18 | Automatic patrol monitoring method and system for unmanned aerial vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116934558A true CN116934558A (en) | 2023-10-24 |
| CN116934558B CN116934558B (en) | 2023-12-19 |
Family
ID=88384742
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311200125.4A Active CN116934558B (en) | 2023-09-18 | 2023-09-18 | Automatic patrol monitoring method and system for unmanned aerial vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116934558B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117664888A (en) * | 2024-01-31 | 2024-03-08 | 北京英视睿达科技股份有限公司 | Water quality monitoring method, device, equipment and medium based on water quality prediction model library |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210374466A1 (en) * | 2020-05-26 | 2021-12-02 | Zhejiang University | Water level monitoring method based on cluster partition and scale recognition |
| US20220309295A1 (en) * | 2021-03-29 | 2022-09-29 | International Business Machines Corporation | Multi-Modal Fusion Techniques Considering Inter-Modality Correlations and Computer Model Uncertainty |
| CN115184352A (en) * | 2022-07-08 | 2022-10-14 | 重庆亿森动力环境科技有限公司 | Water quality monitoring method based on BP neural network |
| CN116543561A (en) * | 2023-07-06 | 2023-08-04 | 之江实验室 | Knowledge and data double-drive-based traffic congestion propagation prediction method |
| CN116593422A (en) * | 2023-04-04 | 2023-08-15 | 绿格兰（天津）环保科技有限公司 | Early warning system for monitoring urban river reach sewage discharge based on infrared remote sensing technology |
-
2023
- 2023-09-18 CN CN202311200125.4A patent/CN116934558B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210374466A1 (en) * | 2020-05-26 | 2021-12-02 | Zhejiang University | Water level monitoring method based on cluster partition and scale recognition |
| US20220309295A1 (en) * | 2021-03-29 | 2022-09-29 | International Business Machines Corporation | Multi-Modal Fusion Techniques Considering Inter-Modality Correlations and Computer Model Uncertainty |
| CN115184352A (en) * | 2022-07-08 | 2022-10-14 | 重庆亿森动力环境科技有限公司 | Water quality monitoring method based on BP neural network |
| CN116593422A (en) * | 2023-04-04 | 2023-08-15 | 绿格兰（天津）环保科技有限公司 | Early warning system for monitoring urban river reach sewage discharge based on infrared remote sensing technology |
| CN116543561A (en) * | 2023-07-06 | 2023-08-04 | 之江实验室 | Knowledge and data double-drive-based traffic congestion propagation prediction method |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117664888A (en) * | 2024-01-31 | 2024-03-08 | 北京英视睿达科技股份有限公司 | Water quality monitoring method, device, equipment and medium based on water quality prediction model library |
| CN117664888B (en) * | 2024-01-31 | 2024-05-03 | 北京英视睿达科技股份有限公司 | Water quality monitoring method, device, equipment and medium based on water quality prediction model library |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116934558B (en) | 2023-12-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111368886B (en) | Sample screening-based label-free vehicle picture classification method | |
| Sathishkumar et al. | Forest fire and smoke detection using deep learning-based learning without forgetting | |
| Li et al. | Semi-supervised semantic segmentation using adversarial learning for pavement crack detection | |
| McIver et al. | Estimating pixel-scale land cover classification confidence using nonparametric machine learning methods | |
| WO2020049098A1 (en) | Computer-implemented method, computer program product and system for analysis of cell images | |
| CN116934558B (en) | Automatic patrol monitoring method and system for unmanned aerial vehicle | |
| CN111178611B (en) | Method for predicting daily electric quantity | |
| CN115223063B (en) | Deep learning-based unmanned aerial vehicle remote sensing wheat new variety lodging area extraction method and system | |
| Duan et al. | A semi-supervised deep learning approach for vessel trajectory classification based on AIS data | |
| CN114694178A (en) | Method and system for monitoring safety helmet in power operation based on fast-RCNN algorithm | |
| CN110675374A (en) | Two-dimensional image sewage flow detection method based on generation countermeasure network | |
| Adegun et al. | Review of deep learning methods for remote sensing satellite images classification: experimental survey and comparative analysis | |
| CN116188756A (en) | Instrument angle correction and indication recognition method based on deep learning | |
| CN115115940A (en) | Blue algae bloom monitoring method and monitoring and early warning system thereof | |
| CN114139624A (en) | Method for mining time series data similarity information based on integrated model | |
| CN114048817A (en) | Deep learning input set priority test method based on variation strategy | |
| JPWO2012032889A1 (en) | Learning apparatus, learning system, learning method, and learning program for object identification | |
| Apeagyei et al. | Evaluation of deep learning models for classification of asphalt pavement distresses | |
| CN116872961B (en) | Control system for intelligent driving vehicle | |
| CN115994327B (en) | Equipment fault diagnosis method and device based on edge calculation | |
| CN116612103A (en) | Intelligent detection method and system for building structure cracks based on machine vision | |
| CN115184054B (en) | Mechanical equipment semi-supervised fault detection and analysis method, device, terminal and medium | |
| CN117169143A (en) | Full spectrum water quality detection method and system based on neural network | |
| CN115376094A (en) | Unmanned sweeper road surface identification method and system based on scale perception neural network | |
| Guillermo et al. | A Comparative Study on Satellite Image Analysis for Road Traffic Detection using YOLOv3-SPP, Keras RetinaNet and Full Convolutional Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |