Cloud and machine combined intelligent air quality decision system and method

Abstract

The invention discloses an intelligent air quality decision-making system and method combined with a cloud aircraft, and the method comprises the steps: designing a self-adaptive objective function, carrying out the optimal path planning of an unmanned aerial vehicle through a moss growth optimization algorithm improved through a chaotic mapping coupling population self-adaptive adjustment strategy, and carrying out the data processing through a multi-source heterogeneous data processing model. Air pollution data and geographic positions collected by the unmanned aerial vehicle in real time serve as input of a BiLSTM-Informer air quality detection model, space-time multi-scale detection is carried out, and auxiliary decision support is provided; the method can better adapt to environment changes, air quality real-time monitoring and intelligent and effective decision making in a severe environment are achieved, and the unmanned aerial vehicle can give consideration to efficiency and safety during path planning through a multi-target optimization method.

Description

Technical Field

[0001] The present invention relates to an intelligent decision-making system and method for air quality, and particularly to an intelligent decision-making system and method for air quality combining cloud and machine. Background Art

[0002] Current air pollution monitoring technologies mainly rely on ground-based fixed monitoring stations, which have problems such as lagging data updates, inability to reflect air quality changes in real time, and difficulty in coping with extremely bad weather. Although remarkable achievements have been made in air quality prediction, such as independently developing air quality models and improving prediction and forecasting capabilities, domestic research methods generally lack pertinence and lack technical specification guidance for parameterization scheme selection, resulting in poor comparability of prediction results and inability to effectively provide accurate and real-time information for environmental protection departments and emergency management departments.

[0003] In view of the above problems, traditional air pollution monitoring means have limitations. The data update cycle of ground monitoring stations is long, and it is difficult to reflect the changing trend of air quality in real time. In addition, environmental factors such as strong winds, heavy rains, and extremely bad weather will affect the sampling and data transmission of ground monitoring points, resulting in difficult data acquisition. Although existing air quality prediction technologies can provide early warning information, due to the lack of localization parameterization of high-quality observation data and the guidance of technical specifications, the model prediction accuracy is limited and it is difficult to meet the application requirements of high precision and rapid response.

[0004] Micro unmanned aerial vehicles, with their characteristics of small size, flexibility, easy deployment, etc., can play an important role in scenarios where traditional monitoring means are incompetent, such as accident site monitoring, high-altitude monitoring, dangerous area monitoring, etc. Combining the computing power and data processing capabilities of the cloud platform, micro unmanned aerial vehicles can achieve real-time monitoring, data transmission, and prediction analysis of air pollutants, providing more accurate, timely, and targeted air pollution prevention and emergency response information for environmental protection departments and emergency management departments. Summary of the Invention

[0005] Object of the Invention: The object of the present invention is to provide an intelligent decision-making system for air quality combining cloud and machine to improve the accuracy of air quality detection and the efficiency and safety of UAV flight path planning. On the other hand, to provide an intelligent decision-making method for air quality combining cloud and machine.

[0006] Technical Solution: The intelligent decision-making system for air quality combining cloud and machine according to the present invention includes:

[0007] An adaptive path planning module, which is used to design an adaptive objective function and perform optimal path planning for the UAV by using the moss growth optimization algorithm improved by the chaotic mapping coupled population adaptive adjustment strategy.

[0008] The multi-source heterogeneous data acquisition module is used to collect and process air pollutant data and aerial image data. The multi-source heterogeneous data acquisition module includes a gas sensor, an ultrasonic sensor, and an accelerometer;

[0009] The central processing unit module is used to receive the data from the multi-source heterogeneous data acquisition module of the drone and transmit the data to the cloud platform using the Wi-Fi communication module;

[0010] The Wi-Fi communication module is used to achieve data transmission between the central processing unit module and the cloud platform;

[0011] The cloud platform is used to provide auxiliary decision-making support. The cloud platform specifically includes:

[0012] The data management module is used to clean and integrate the collected data, remove the duplicate and incorrect data therein to improve the accuracy of the final result, integrate the time series monitored by different sensors and the aerial image data of the drone for subsequent retrieval and analysis, and regularly back up the data to ensure that the system can recover the data in case of data loss or damage;

[0013] The air quality prediction module is used to drive the BiLSTM-Informer model to predict the future air quality according to the current air quality status;

[0014] The performance evaluation module is used to evaluate the performance of the gas sensor carried by the drone;

[0015] The cloud platform display module is used to receive and display the prediction results of the future air quality;

[0016] The air quality remote control module is used to remotely control the flight path and monitoring tasks of the drone, achieve flexible coverage of the monitoring area, evaluate the environmental quality, and issue a warning when the pollutant exceeds the standard.

[0017] Preferably, the calculation formula of the adaptive objective function in the adaptive path planning module is:

[0018] f min =(1 - p 1 - p 2 )f 1 + p 1 f 2 + p 2 f 3 ;

[0019]

[0020] Among them, f 1 represents the voyage cost, f 2 represents the threat cost, f 3 represents the altitude cost, p1 and p 2 represents the importance degree coefficient of the objective function, and p 2 is dynamically adjusted according to the data collected by the ultrasonic sensor;

[0021] λ 1 and λ 2 represent the cost weight coefficient of the power consumption during the flight time of the UAV. t(n, n + 1) represents the flight time of the UAV from the nth flight path point to the (n + 1)th flight path point, q represents the proportionality coefficient of the unit power consumption being proportional to the flight distance, represents the total flight distance between any two points in the three-dimensional search space of the UAV;

[0022] d n represents the danger degree of the nth flight node, sum_obstacle represents the number of obstacle grids around the flight path point, and sum_surround represents the total number of grids around the flight path point;

[0023] n num represents the total number of flight path points, and φ max and φ min represent the weight coefficients for exceeding the maximum height and being lower than the minimum height, and φ rate represents the weight coefficient of the height change rate, and h max and h min represent the set maximum and minimum flight heights, and a n 、b n 、c n represent the nth flight path node of the UAV flight path;

[0024] θ max represents the maximum pitch angle of the UAV, and β max represents the maximum turning angle.

[0025] Preferably, the BiLSTM-Informer model formula in the air quality prediction module is:

[0026]

[0027] Q m = MultiHeadAttention(X′ t );

[0028] C m = Softmax{ReLU|(W q Q m ) T W O Q m}};

[0029]

[0030] Among them, X t represents the input of the BiLSTM-Informer model, p represents the spatial sampling points, m represents the types of air pollutants, and n represents the sample size of the concentration of the m-th air pollutant;

[0031] Q represents the query matrix, O represents the key-value matrix, L represents the value matrix, and W q and W O and W L represent trainable weight matrices, d k represents the number of columns of the Q, O, and L matrices, that is, the vector dimension. Q m represents the local features in the head attention mechanism, and X' t represents the different time series segments into which the input time series data is decomposed, and C m represents the feature weights of different time series sequences, and S (m) represents the self-attention weights between each time series segment, represents the output pollutant concentration at time t, and U m represents the hyperparameter matrix in the BiLSTM-Informer model, which can be optimized to improve the model accuracy and generalization ability. Softmax is a function in deep learning that can convert the original output of the model into a probability distribution. MultiHeadAttention is the multi-head self-attention mechanism for processing sequence data.

[0032] Preferably, the performance in the performance evaluation module includes:

[0033] The precision, sensitivity, response time, and stability for evaluating the data collection ability of the drone;

[0034] The data collection frequency, data real-time performance, and data processing efficiency for evaluating the data processing ability of the drone transmission.

[0035] An intelligent air quality decision-making method combining cloud and machine according to the present invention includes the following steps:

[0036] (1) Design the adaptive objective function of the drone through the adaptive path planning module, and use the moss growth optimization algorithm improved by the chaos mapping coupled population adaptive adjustment strategy to perform optimal path planning for the drone;

[0037] (2) Collect air pollutant data and aerial image data through the multi-source heterogeneous data processing module and process them;

[0038] (3) The central processing unit module receives the data from the drone multi-source heterogeneous data collection module and transmits the data to the cloud platform through the Wi-Fi communication module;

[0039] (4) The cloud platform uses the BiLSTM-Informer prediction model to predict the future air quality based on the current air quality status, evaluate the performance of the gas sensors carried by the drone, display the final prediction results of the future air quality, and provide auxiliary decision-making support;

[0040] (5) Real-time monitor data through the air quality remote control module. When the detected pollutant concentration exceeds the preset threshold, trigger the warning mechanism and display the pollutant hotspots and warning areas.

[0041] Preferably, the optimal path planning for the drone in step 1 is as follows:

[0042] (11) Initialize the drone position using the chaotic map and determine the optimal path search direction according to the adaptive objective function value;

[0043] (12) Update the drone position according to the search direction and search step size;

[0044] (13) Locally update the drone position to achieve the optimal path planning of the drone.

[0045] Preferably, the chaotic map formula in step 1 is:

[0046]

[0047] where R represents the algorithm search space variable, r represents the control parameter, R ∈ [0, 1], r ∈ (0, 4), M(0) represents the optimized initial position of the drone, α and β represent the lower and upper limits of the algorithm search range, and mod 1 represents the modulo operation with respect to 1;

[0048] The formula for the population adaptive adjustment strategy is:

[0049]

[0050] where M bad represents the worst drone position, M bes t represents the best drone position, v represents the current objective function evaluation, % represents the remainder or modulo operator, FE s represents the current iteration number, and MaxFE s represents the maximum iteration number.

[0051] Preferably, the formula for determining the optimal path search direction in step 11 is:

[0052]

[0053] where f represents the adaptive objective function, D dirrepresents the calculated search direction, and num represents the total number of alternative paths of the UAVs in divV.

[0054] Preferably, the formula for updating the UAV position according to the search direction and search step in step 22 is:

[0055]

[0056] where, M i represents the current position of the UAV, D dir represents the calculated search direction, and the search step is determined according to the relative magnitudes of the random number p 1 and d 1 ;

[0057] When p 1 > d 1 , long - distance search is adopted: F 1 = w·(p 2 - 0.5)·E;

[0058] where, w represents a constant parameter, w is set to 2, p 2 represents a random variable within the range of (0, 1), and E is the standard search step;

[0059] When p 1 ≤ d 1 , short - distance search is adopted:

[0060]

[0061] where, p 3 is a random vector within the range of (0, 1) and can control the search step;

[0062] FE s represents the current iteration number, and MaxFE s represents the maximum iteration number.

[0063] Preferably, the formula for locally updating the UAV position in step 13 is:

[0064]

[0065] where, represents the position of the i - th UAV, d 2 and p 4 represent a random number between (0, 1);

[0066] It is judged whether to update the UAV position using M best through the position update factor act, and the calculation formula is:

[0067]

[0068] F 3 = 0.1·(p 6 - 0.5)·E;

[0069] where p 5 represents a random vector within the range (0, 1), and p 6 represents a random number within the range (0, 1).

[0070] Advantages: Compared with the prior art, the present invention has the following remarkable advantages: 1. By considering three important factors of voyage cost, threat cost, and altitude cost, the flexibility and adaptability of the algorithm are improved; 2. The moss growth optimization algorithm improved by using the chaotic mapping coupled population adaptive adjustment strategy ensures the uniform distribution of unmanned aerial vehicles, thereby improving work efficiency and the accuracy of data collection; 3. By constructing a BiLSTM-Informer prediction model, the accuracy of prediction is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 is a schematic diagram of the system framework of the present invention;

[0072] Figure 2 is a schematic diagram of the optimal path planning of the present invention;

[0073] Figure 3 is a schematic diagram of the air quality prediction model of the present invention;

[0074] Figure 4 is a schematic diagram of the BiLSTM-Informer air quality prediction model of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0075] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0076] (1) Through the adaptive path planning module, the flight altitude of the unmanned aerial vehicle is dynamically guided according to the ultrasonic sensor data, and an adaptive objective function that simultaneously considers the safety, flight efficiency, and task completion degree of the unmanned aerial vehicle is designed to perform optimal trajectory planning for the unmanned aerial vehicle and realize dynamic adjustment of the trajectory. The calculation formula of the adaptive objective function is as follows:

[0077] f min = (1 - p 1 - p 2 )f 1 + p 1 f 2 + p 2 f 3 ;

[0078]

[0079]

[0080] Among them, f 1 represents the voyage cost, and f 1 emphasizes efficiency and tries to shorten the flight distance as much as possible. f 2 represents the threat cost, and f 3 represents the altitude cost, and f 2 and f 3 focus on safety and require the UAV to avoid threat areas and maintain a safe flight altitude. p 1 and p 2 represent the importance degree coefficients of the objective function. p 2 Dynamically adjust according to the data collected by the ultrasonic sensor, so as to dynamically adjust the weights of the other two cost functions, realize the dynamic adjustment of the weights of these three costs, balance the three, and ensure the efficient and safe execution of the task;

[0081] λ 1 and λ 2 represent the cost weight coefficients of the power consumption of the UAV flight time. t(n, n + 1) represents the flight time of the UAV from the nth flight path point to the (n + 1)th flight path point, and q represents the proportional coefficient that the unit power consumption is proportional to the flight distance. represents the total flight distance between any two points in the three-dimensional search space of the UAV;

[0082] d n represents the danger degree of the nth flight node, sum_obstacle represents the number of obstacle grids around the flight path point, sum_surround represents the total number of grids around the flight path point, and n num represents the total number of flight path points;

[0083] φ max and φ min represent the weight coefficients for exceeding the maximum altitude and being lower than the minimum altitude. φ rate represents the weight coefficient of the altitude change rate, h max and h min represent the set maximum and minimum flight altitudes, a n 、b n 、c n represent the nth flight path node of the UAV flight path, a n+1 、b n+1 、c n+1 represent the (n + 1)th flight path node of the UAV flight path, θ max represents the maximum pitch angle of the UAV, and β max represents the maximum turning angle.

[0084] (2) The moss growth optimization algorithm improved by using the chaos mapping coupled population adaptive adjustment strategy is used to perform optimal path planning for the UAV.

[0085] 21. Initialize the UAV position M using the chaos mapping to ensure that more uniform UAV positions are generated in the search space, avoid premature convergence of the optimization process to the local optimum, thereby improving the global search ability, and also help the optimization process better adapt to environmental changes. The mathematical formula of the chaos mapping is expressed as follows:

[0086]

[0087] Among them, R represents the algorithm search space variable, r represents the control parameter, R ∈ [0, 1], r ∈ (0, 4), M(0) represents the optimized initial position of the UAV, α and β represent the lower and upper limits of the algorithm search range, and mod 1 represents the modulo operation with respect to 1. Since the chaos mapping is performed within the interval [0, 1], it is necessary to ensure that the search results are always within this interval.

[0088] Determine the optimal path search direction D according to the adaptive objective function value dir :

[0089]

[0090] Among them, f represents the adaptive objective function, D dir represents the calculated search direction, and num represents the total number of alternative paths of the UAV in divV.

[0091] 22. Update the UAV position according to the search direction and search step size. After the position update, a new UAV position is formed. The wind intensity decreases as the iteration progresses to simulate the process of the UAV gradually adapting to the environment and finding a better path. In this way, the UAV individuals can better adapt to environmental changes and improve the efficiency and accuracy of their path planning. The updated position formula is as follows:

[0092]

[0093] Among them, D dir represents the search direction, M i represents the current position of the UAV, and the search step size is determined according to the relative sizes of the random numbers p 1 and d 1 ;

[0094] When p 1 > d 1 , long-distance search is adopted:

[0095] F 1 = w·(p 2 - 0.5)·F;

[0096] Among them, w represents a constant parameter, and w is set to 2, p 2 represents a random variable within the range (0, 1);

[0097] When p 1 ≤ d 1 At this time, short-distance search is adopted:

[0098]

[0099] Among them, p 3 represents a random vector within the range (0, 1), which can control the search step size;

[0100]

[0101] E is the standard search step size, FE s represents the current iteration number, MaxFE s represents the maximum iteration number.

[0102] Adopt a population adaptive adjustment strategy to balance global search and local search by dynamically updating the moss positions during the optimization process. This dynamic adjustment helps to ensure that the UAV can find the optimal path while accelerating the search efficiency. The method is as follows:

[0103]

[0104] Among them, M bad represents the worst UAV position, M best represents the best UAV position, v represents the current objective function evaluation, % represents the remainder or modulo operator, FE s represents the current iteration number, MaxFE s represents the maximum iteration number.

[0105] 23. Locally update the UAV position to achieve the planning of the UAV's optimal path

[0106] Make the UAV pay more attention to the area near the current optimal path during the path planning process, so as to find more possible better paths, thereby enhancing the UAV's local area exploration ability. The formula is as follows:

[0107]

[0108] Among them, represents the position of the i-th UAV, d 2 and p 4 are a random number between (0, 1);

[0109] The position update factor act judges whether to use Mbest Update the position of the drone, and the calculation formula is as follows:

[0110]

[0111] F 3 = 0.1·(p 6 - 0.5)·E;

[0112] where p 5 represents a random vector within the range (0, 1), and p 6 represents a random number within the range (0, 1).

[0113] (3) Process the time series data and aerial image data obtained by the pollutant detection sensor.

[0114] In the multi-source heterogeneous data acquisition module, process the time series data and aerial image data of the pollutant detection sensor equipped on the drone. Among them, the gas sensor is used to detect harmful gases in the air; the accelerometer provides the acceleration forces borne by the drone in the x, y, and z axis directions, determines the tilt angle of the drone when it is in a stationary state, and provides linear acceleration data in the horizontal and vertical directions; the ultrasonic sensor uses the characteristic that ultrasonic waves will bounce back when they encounter other substances to control the height. Especially when near the ground, it is used in combination with the pressure sensor to achieve stable flight of the drone at high altitudes and low altitudes.

[0115] (4) Take the air pollutant data and geographical location collected by the drone as input data, and input the input data into the BiLSTM-Informer air quality detection model.

[0116] During the path optimization process, the historical positions of each drone path planning will be recorded. By selecting the best drone position in the historical records, the initial drone position is updated, which can avoid repeated searches and improve search efficiency. Construct the BiLSTM-Informer air quality detection model, and the formula is as follows:

[0117]

[0118] Q m = MultiHeadAttention(X′ t );

[0119] C m = Softmax{ReLU(W q Q m ) T W O Q m};

[0120]

[0121] Among them, i t represents the input gate output, σ represents the sigmod activation function, W xi represents the input weight of the input gate, x t represents the input at time t, W hi represents the weight between the hidden state and the input gate, h t-1 represents the hidden state at time t-1, b i represents the bias between the hidden state and the input gate, f t represents the output of the forget gate, W xf represents the weight between the output gate and the forget gate, b f represents the bias between the hidden state and the forget gate, represents the output of the neuron, W xc represents the bias between the input gate and the neuron, W hc represents the weight between the hidden state and the neuron, b c represents the bias between the hidden state and the neuron, C t represents the cell state at time t, o t represents the output of the output gate, W xo represents the weight between the input gate and the output gate, W ho represents the weight between the hidden state and the output gate, b ho represents the bias between the hidden state and the output gate;

[0122] X t represents the input of the BiLSTM-Informer model, p represents the spatial sampling points, m represents the types of air pollutants, n represents the sample size of the concentration of the m-th air pollutant, Q represents the query matrix, O represents the key matrix, L represents the value matrix, W q 、W O 、W L represents the trainable weight matrix, d k represents the number of columns of the Q, O, L matrices, that is, the vector dimension, Q m represents the local features in the head attention mechanism, X' t represents the different time series segments into which the input time series data is decomposed, C m represents the feature weights of different time series sequences, S (m) represents the self-attention weight between each time series segment, represents the output pollutant concentration at time t, U mIt represents the hyperparameter matrix in the BiLSTM-Informer model, which can be optimized to improve the model's accuracy and generalization ability. Softmax is a function in deep learning that can convert the original output of the model into a probability distribution. MultiHeadAttention is a multi-head self-attention mechanism used to process sequence data.

[0123] The air quality prediction module uses the BiLSTM-Informer model, which consists of an encoder, a decoder, and an output layer.

[0124] The sparse multi-head probabilistic sub-attention mechanism and distillation operation in the encoder are key technologies for improving efficiency and performance. The sparse multi-head probabilistic sub-attention mechanism is an optimized attention mechanism that improves the model's efficiency by reducing unnecessary computations, especially when dealing with long sequences. Its core lies in only focusing on the most important parts of the sequence, reducing the computational amount, and mapping the input features to different subspaces, enabling the model to simultaneously focus on information from different representation subspaces and enhancing the model's ability to capture information. The distillation operation is a model compression technology that realizes by transferring the information of a complex model to a simple model.

[0125] In the decoder of the BiLSTM-Informer model, BiLSTM is used to process sequence data through two LSTM networks, forward and backward. This enables the model to simultaneously obtain information before and after each time point in the sequence, enhancing the understanding and prediction ability of time series data. The sparse multi-head probabilistic self-attention mechanism reduces the computational amount by introducing a mask, helping the model focus resources on processing key information. The multi-head self-attention mechanism allows the model to capture information in different representation subspaces. The weights learned by each "head" represent the importance of different features, and these weights can be regarded as probability distributions, guiding the model to focus resources on processing the most relevant information.

[0126] In the output layer, the linear layer and the softmax function are two key components of the output layer. They jointly are responsible for converting the internal representation of the model into the final classification result. The linear layer is used to receive the output of the BiLSTM layer and map the features it extracts to the final prediction space. The softmax function is usually applied after the fully connected layer. The fully connected layer fuses and transforms the output features of BiLSTM and Informer, and then the softmax function converts these transformed features into the results of the final air quality prediction module. Use MAE, RMSE, and R 2As an evaluation metric for the model, it is used to quantify the difference between the model's predicted values and the actual observed values, thereby evaluating the accuracy of the model. The performance evaluation metric is not only used to evaluate the model's performance, but also to identify the key factors affecting the model's performance and adjust the model accordingly to improve the prediction accuracy.

[0127] (5) The cloud platform monitors the data of air quality sensors and monitoring stations in real time. Once it detects that the pollutant concentration exceeds the preset threshold, it immediately triggers the warning mechanism. According to the air quality index or the concentration of specific pollutants, three warning levels are set: yellow warning, orange warning, and red warning. After the warning is issued, the information will be automatically sent to users as air quality warnings via email, text message, mobile app push, etc. On the user interface of the cloud platform, the pollution hotspots and warning areas will be visually displayed through visualization tools such as maps, charts, and dashboards, which is convenient for users to plan future travel modes or routes in advance and avoid the harm of air pollutants to themselves. When emergencies occur, such as gas leaks, the emergency management department can take emergency measures, formulate corresponding evacuation routes, provide auxiliary decision-making support, and ensure the safety of personnel.

Claims

1. A cloud-machine-integrated air quality intelligent decision-making system, characterized in that: include: The adaptive path planning module is used to design an adaptive objective function and use the moss growth optimization algorithm improved by the chaotic map coupled population adaptive adjustment strategy to plan the optimal path for the UAV; A multi-source heterogeneous data acquisition module, used to collect and process air pollutant data and aerial image data, wherein the multi-source heterogeneous data acquisition module includes a gas sensor, an ultrasonic sensor, and an accelerometer; The central processing unit module is used to receive data from the multi-source heterogeneous data acquisition module of the drone and transmit the data to the cloud platform using the wireless fidelity communication module; Wireless fidelity communication module, used to realize data transmission between the central processing unit module and the cloud platform; A cloud platform is used to provide auxiliary decision support, and the cloud platform specifically includes: Data management module, used to clean and integrate collected data; The air quality prediction module is used to drive the BiLSTM-Informer model to predict future air quality based on current air quality conditions; Performance evaluation module, used to evaluate the performance of the gas sensor carried by the drone; A cloud platform display module is used to receive and display the prediction results of future air quality; The air quality remote control module is used to remotely control the flight path and monitoring tasks of the drone, evaluate environmental quality, and issue warnings when pollutants exceed the standard.

2. The cloud-machine-integrated air quality intelligent decision-making system according to claim 1 is characterized in that: The calculation formula of the adaptive objective function in the adaptive path planning module is: f min =(1-p1-p2)f1+p1f2+p2f3; Among them, f1 represents the range cost, f2 represents the threat cost, f3 represents the altitude cost, p1 and p2 represent the importance coefficients of the objective function, and p2 is dynamically adjusted according to the data collected by the ultrasonic sensor; λ1 and λ2 represent the cost weight coefficients of the power consumption of the UAV during flight time, t(n,n+1) represents the flight time of the UAV from the nth track point to the n+1th track point, and q represents the proportional coefficient of the unit power consumption proportional to the flight distance. It represents the total flight distance between any two points of the UAV in the three-dimensional search space; d n Indicates the danger level of the nth flight node, sum_obstacle indicates the number of obstacle grids around the track point, and sum_surround indicates the total number of grids around the track point; n num Represents the total number of track points, φ max and φ min Represents the weight coefficient for exceeding the maximum height and falling below the minimum height, φ rate The weight coefficient representing the rate of change of height, h max and h min Indicates the maximum and minimum flight altitudes set, a n , b n 、c n Indicates the nth track node of the UAV flight aircraft; θ max Indicates the maximum pitch angle of the drone, β max Indicates the maximum turning angle.

3. The cloud-machine-integrated air quality intelligent decision-making method according to claim 1 is characterized in that: The BiLSTM-Informer model formula in the air quality prediction module is: Q m =MultiHeadAttention(X′ t ); Among them, X t represents the input of the BiLSTM-Informer model, p represents the spatial sampling point, m represents the type of air pollutant, and n represents the sample size of the mth air pollutant concentration; Q represents the query matrix, O represents the key-value matrix, L represents the value matrix, and W q , W O , W L represents the trainable weight matrix, d k Indicates the number of columns of the Q, O, and L matrices, that is, the vector dimension, Q m represents the local features in the head attention mechanism, X' t Indicates the different time series segments into which the input time series data is decomposed, C m Represents the feature weights of different time series, S (m) represents the self-attention weight between each temporal segment, represents the output pollutant concentration at time t, U m Represents the hyperparameter matrix in the BiLSTM-Informer model.

4. The cloud-machine-integrated air quality intelligent decision-making system according to claim 1 is characterized in that: The performance in the performance evaluation module includes: Used to evaluate the accuracy, sensitivity, response time and stability of the drone's data collection capabilities; It is used to evaluate the data collection frequency, data real-time and data processing efficiency of the UAV's transmission data processing capabilities.

5. A cloud-machine-integrated intelligent air quality decision-making method, characterized in that: The following steps are involved: (1) The adaptive objective function of the UAV is designed through the adaptive path planning module, and the optimal path planning of the UAV is carried out by using the moss growth optimization algorithm improved by the chaotic mapping coupled population adaptive adjustment strategy; (2) Collect and process air pollutant data and aerial image data through a multi-source heterogeneous data processing module; (3) The central processing unit module receives data from the multi-source heterogeneous data acquisition module of the drone and transmits the data to the cloud platform using the wireless fidelity communication module; (4) The cloud platform uses the BiLSTM-Informer prediction model to predict future air quality based on current air quality conditions, evaluate the performance of the gas sensors carried by the drone, display the final prediction results of future air quality, and provide auxiliary decision support; (5) The air quality remote control module monitors data in real time. When the pollutant concentration exceeds the preset threshold, the early warning mechanism is triggered to display the pollutant hotspots and early warning areas.

6. The cloud-machine combined air quality decision-making method according to claim 5 is characterized in that: Step 1 is to plan the optimal path for the drone. The specific steps are as follows: (11) Using chaotic mapping to initialize the UAV position, and determining the optimal path search direction according to the adaptive objective function value; (12) Update the drone position according to the search direction and search step size; (13) Locally update the UAV position to achieve the optimal path planning for the UAV.

7. The cloud-machine combined air quality decision-making method according to claim 5 is characterized in that: The chaotic mapping formula in step 1 is: Where R represents the algorithm search space variable, r represents the control parameter, R∈[0,1], r∈(0,4), M(0) represents the optimized initial position of the UAV, α and β represent the lower and upper limits of the algorithm search range, and mod 1 represents the modulo operation of 1; The population adaptive adjustment strategy formula is: Among them, M bad represents the worst drone position, M best represents the optimal drone position, v represents the current objective function evaluation, % represents the remainder or modulus operator, FE s Indicates the current number of iterations, MaxFE s Indicates the maximum number of iterations.

8. The cloud-machine combined air quality decision-making method according to claim 6 is characterized in that: The formula for determining the optimal path search direction in step 11 is: Where f represents the adaptive objective function, D dir Represents the calculated search direction, and num represents the total number of alternative paths for the drone in divV.

9. The cloud-machine combined air quality decision-making method according to claim 6 is characterized in that: The formula for updating the drone position according to the search direction and search step length in step 22 is: Among them, M i Indicates the current position of the drone, D dir represents the calculated search direction, and the search step size is determined by the relative size of the random numbers p1 and d1; When p1>d1, long distance search is adopted: F1=w·(p2-0.5)·E; Where w represents a constant parameter, w is set to 2, p2 represents a random variable in the range of (0,1), and E is the standard search step size; When p1≤d1, short distance search is used: Among them, p3 is a random vector in the range of (0,1) that can control the search step size; FE s Indicates the current number of iterations, MaxFE s Indicates the maximum number of iterations.

10. The cloud-machine combined air quality decision-making method according to claim 6 is characterized in that: The formula for locally updating the drone position in step 13 is: in, represents the position of the i-th drone, d2 and p4 represent a random number between (0,1); Determine whether to use M by using the location update factor act best Update the drone position, the calculation formula is: F3 = 0.1·(p6-0.5)·E; Among them, p5 represents a random vector in the range of (0,1), and p6 represents a random number in the range of (0,1).

Images