An agricultural facility monitoring system based on adaptive wireless communication

Abstract

The present invention discloses a facility agriculture monitoring system based on adaptive wireless communication, which relates to the field of facility agriculture monitoring. It includes a monitoring and sensing module, a wireless communication module, a processing and scheduling module, a power management module, and a real-time monitoring module. The improved SMDP reinforcement decision algorithm is used to adaptively adjust the communication rate and communication protocol, and a node multi-hop communication path and a multi-path transmission mechanism are established to cope with communication interference and faults in the wireless communication network. An adaptive scheduling strategy model is used to dynamically adjust the monitoring tasks and data acquisition frequencies of sensor nodes, and an energy management model and a sleep-wakeup mechanism are used to dynamically adjust the power supply strategy. The problem solved is that the traditional facility agriculture monitoring system cannot adapt to the dynamic changes of agricultural production. The present invention can automatically adjust and optimize according to the results of real-time environmental monitoring, improving the accuracy and timeliness of monitoring data.

Description

Technical Field

[0001] The present invention relates to the field of facility agriculture monitoring, and particularly to a facility agriculture monitoring system based on adaptive wireless communication. Background Art

[0002] With the advancement of agricultural modernization, facility agriculture monitoring has become an important means to improve crop yield and quality. Traditional farmland monitoring methods have problems such as high labor costs and low data collection efficiency, and can no longer meet the development needs of modern agriculture. Therefore, there is an urgent need for an efficient and intelligent monitoring technology to replace traditional methods. The rapid development of wireless sensor network technology provides a new solution for facility agriculture monitoring. Wireless sensor nodes are deployed in the monitoring area and communicate wirelessly to achieve real-time monitoring and data transmission of environmental information. In recent years, wireless sensor networks have shown great application potential in agricultural environment monitoring.

[0003] However, the agricultural environment is complex and diverse, such as temperature, humidity, and light. Traditional wireless sensor networks for agricultural monitoring often adopt fixed sensor node collection frequencies, power supply methods, communication protocols, and communication rates, and cannot adapt to the dynamic changes of agricultural production. Therefore, they are greatly affected by agricultural environment interference and have poor adaptability and stability.

[0004] Therefore, a facility agriculture monitoring system based on adaptive wireless communication is needed to solve the above problems. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention discloses a facility agriculture monitoring system based on adaptive wireless communication, which can automatically adjust and optimize according to the results of real-time environmental monitoring, improve the accuracy and timeliness of monitoring data, and has a high degree of automation and intelligence.

[0006] The present invention adopts the following technical solutions:

[0007] A facility agriculture monitoring system based on adaptive wireless communication, comprising:

[0008] A monitoring and sensing module, the monitoring and sensing module uses multiple sensor nodes to collect environmental parameters in the facility agriculture monitoring area. The sensor nodes are built-in with multiple types of low-power sensing chip groups. The monitoring and sensing module cross-verifies and compares the data collected by different types of sensing chips through a cooperation mechanism to remove error data;

[0009] A wireless communication module that uses a wireless communication network to achieve data communication and information exchange between different sensor nodes, and transmits and receives the data collected by different sensor nodes to a cloud convergence node for further processing. The wireless communication module adaptively adjusts the communication rate and communication protocol of the wireless communication network using an improved SMDP reinforcement decision algorithm, and responds to communication interference and faults in the wireless communication network by establishing a node multi-hop communication path and a multi-path transmission mechanism;

[0010] A processing and scheduling module that uses an adaptive scheduling strategy model to schedule the monitoring tasks of multiple sensor nodes. The adaptive scheduling strategy model dynamically adjusts the monitoring tasks and data collection frequencies of the sensor nodes according to the needs of agricultural production and the monitoring results of facility agricultural environment parameters;

[0011] A power management module that dynamically adjusts the power supply strategy using an energy management model. The energy management model manages the energy consumption of sensor nodes through a sleep-wake mechanism and dynamically adjusts the power output according to the energy consumption requirements of sensor nodes, the wireless communication module, and the processing and scheduling module;

[0012] A real-time monitoring module that performs real-time data monitoring, historical data query, and remote control on environmental parameters and wireless communication status parameters in the monitoring area of facility agriculture through a visualization platform, and uses a user interface to implement the interactive operation between the user and the visualization platform.

[0013] Further, the multi-type low-power sensing chipset includes at least a temperature sensing chip, a humidity sensing chip, a soil humidity sensing chip, a light sensing chip, a carbon dioxide sensing chip, and a conductivity sensing chip.

[0014] Further, the improved SMDP reinforcement decision algorithm dynamically adjusts the communication strategy according to communication distance, data volume, network load, and channel quality. The working method of the improved SMDP reinforcement decision algorithm includes the following steps:

[0015] S1. Define the state parameters and communication strategies of the wireless communication network, and initialize the calculation parameters. The state parameters include at least channel quality, network topology, transmission rate, and protocol type. The calculation parameters include learning rate, behavior evaluation value function, discount factor, and policy parameters;

[0016] S2. Add reward and penalty terms to the state utility function according to the requirements of facility agriculture monitoring and the performance indicators of the wireless communication network to evaluate the advantages and disadvantages of different communication strategies. When the data transmission success rate of the wireless communication network is less than the preset threshold, add a reward term; when the time taken for the data to be successfully transmitted to the aggregation node is less than the preset threshold, add a reward term; when the number of sensor nodes simultaneously transmitting data in the current wireless communication network is greater than that in the previous moment, add a reward term; otherwise, add a penalty term.

[0017] S3. Iteratively optimize the communication strategy based on the behavior evaluation value. Select a communication protocol and communication rate from the communication protocol library of the wireless communication network according to the current communication strategy and execute the action. Record the state transition, cumulative reward or penalty, and the state after the transition of the wireless communication network during the execution of the action, and calculate the behavior evaluation value based on the Bellman equation and the recorded data. Sort the expected utility values of the communication strategies and select the communication strategy with the highest expected utility value as the currently selected strategy.

[0018] S4. Execute the data transmission operation according to the currently selected communication strategy, and update the state parameters and calculation parameters according to the transmission result. Record the current state and the corresponding behavior evaluation value in the historical state list.

[0019] S5. Train according to the historical state list and the behavior evaluation value function, and continuously adjust the value of the behavior evaluation value function and the selection of the communication strategy.

[0020] S6. During the communication process, dynamically adjust the communication strategy according to factors such as communication distance, data volume, network load, and channel quality. Repeat steps S2 - S6 until the preset end condition is reached.

[0021] Further, the communication protocol library at least includes the Zigbee short - range wireless sensor network protocol, the LoRa long - range low - power wireless communication protocol, the Wi - Fi wireless local area network protocol, and the NB - IoT narrow - band Internet of Things protocol.

[0022] Further, the multi - hop communication path of the nodes realizes cross - node data transmission through data relay transmission between multiple nodes. The multi - path transmission mechanism divides the data packet to be transmitted into multiple data blocks, transmits them through different transmission paths, and uses a standby drive card to realize path switching when a transmission path fails.

[0023] Further, the standby drive card includes a high - speed serial expansion bus PCIe and a standby control drive circuit. The high - speed serial expansion bus PCIe uses a QOS anti - delay blocking service engine to realize end - to - end communication between the failed transmission path and the standby transmission path.

[0024] Furthermore, the adaptive scheduling strategy model includes an input layer, a data layer, a model layer, an algorithm layer, an optimization layer, and an output layer. The working method of the adaptive scheduling strategy model includes the following steps:

[0025] Step 1: Receive the user's requirements and the environmental parameters of the facility agriculture monitoring area through the input layer, and perform format conversion. Then, transfer the data to the data layer for further processing. The environmental parameters at least include temperature, humidity, light intensity, and carbon dioxide concentration.

[0026] Step 2: The data layer preprocesses and cleans the data transferred from the input layer, and uses the principal component analysis method to integrate the data from different sensor nodes to generate a data set.

[0027] Step 3: The model layer obtains the limiting conditions from the generated data set, and establishes an optimal scheduling mathematical model based on the limiting conditions. The limiting conditions include the calculation scale, the objective function, the constraint conditions, and the variable range.

[0028] Step 4: The algorithm layer classifies and predicts the data set based on the time series prediction algorithm to discover and mine the potential laws and patterns in the data, and dynamically adjusts the monitoring tasks and data collection frequencies of the sensor nodes according to the classification and prediction results.

[0029] Step 5: Through the optimization layer, perform iterative calculations, parameter corrections, and compare the calculation results with the real values, and set the threshold and the number of iterations through the adaptive parameter selection method. The optimization layer uses the parallel calculation method to distribute the calculation tasks to multiple processors or computing nodes to improve the calculation speed.

[0030] Step 6: The output layer transfers the model training results and the task scheduling and data collection frequencies of the nodes to the processing and scheduling module to achieve the dynamic scheduling and task allocation of the sensor nodes.

[0031] Furthermore, the time series prediction algorithm predicts the change of the environmental parameters at the next moment according to the environmental parameters at the historical moment. The data set of the environmental parameter collection at the historical moment is , and the data set of the environmental parameter characteristics affecting the monitoring tasks of the sensor nodes is . The predicted output function of the environmental parameter characteristics affecting the monitoring tasks of the sensor nodes at the t+1 moment is:

[0032]

[0033] In the formula, is the environmental parameter characteristic affecting the monitoring tasks of the sensor nodes at the t+1 moment, It is a weighted function for predicting the characteristics of environmental parameters that affect the monitoring tasks of sensor nodes, and is used to adjust the influence degree of different historical data points on the predicted value. It is the characteristic of environmental parameters that affects the monitoring tasks of sensor nodes at time t. It is the characteristic of environmental parameters that affects the monitoring tasks of sensor nodes at time t - 1. It is the environmental parameter acquisition data at time t. It is the environmental parameter acquisition data at time t - 1. It is the ordinal number of the environmental parameter acquisition time. 。

[0034] Furthermore, the energy management model uses edge computing to compare the current working state and power usage of the sensor node with preset values to obtain the working intensity of the sensor node, and matches the power working mode based on the working intensity. The sleep-wake mechanism automatically adjusts the sleep and wake cycles of the sensor node according to the monitoring requirements and the status of the sensor node.

[0035] The beneficial effects of the present invention are as follows:

[0036] 1. By monitoring the multi-type low-power sensor chipset in the monitoring sensing module, the present invention can comprehensively collect various environmental parameters in the monitoring area of facility agriculture, and apply a collaborative mechanism to cross-verify and compare the data collected by different types of sensor chips, effectively removing incorrect data and improving the accuracy and reliability of the data.

[0037] 2. The wireless communication module of the present invention adopts an improved SMDP reinforcement decision algorithm, which can adaptively adjust the communication rate and communication protocol according to the network conditions, optimize the data transmission efficiency, and establish a node multi-hop communication path and a multi-path transmission mechanism, effectively coping with communication interference and faults in the wireless communication network and enhancing the stability and reliability of the network.

[0038] 3. The present invention adopts an adaptive scheduling strategy model, which can dynamically adjust the monitoring tasks and data acquisition frequency of the sensor node according to the needs of agricultural production and the monitoring results of the agricultural environment, thereby optimizing resource utilization, ensuring that the sensor node can efficiently complete the monitoring tasks, reducing unnecessary energy consumption at the same time, and adopting an energy management model, which can dynamically adjust the power supply strategy to meet the needs of different monitoring tasks. The application of the sleep-wake mechanism effectively manages the energy consumption of the sensor node and extends the overall service life of the network. Description of the Drawings

[0039] Figure 1 It is a schematic diagram of the overall architecture of the present invention;

[0040] Figure 2Schematic diagram of the working process of the improved SMDP reinforcement decision algorithm in the present invention;

[0041] Figure 3 Schematic diagram of the working process of the adaptive scheduling strategy model in the present invention. Detailed implementation manners

[0042] Next, the technical solutions in the embodiments of the present invention will be described clearly and completely in conjunction with the accompanying Figure 1 drawings to Figure 3 Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0043] An embodiment of the present invention discloses a facility agriculture monitoring system based on adaptive wireless communication. As shown in the accompanying Figure 1 drawings, it includes:

[0044] A monitoring and sensing module, including multiple sensor nodes. Each node is built-in with a multi-type low-power sensing chipset, and can achieve precise monitoring by reading multi-dimensional environmental parameters (such as temperature, humidity, light, soil temperature, soil humidity, soil acidity, etc.). A cooperation mechanism is adopted inside the sensor nodes to cross-verify and compare the data collected by different types of sensing chips to improve the accuracy of the data.

[0045] A wireless communication module, which supports wireless communication and message exchange between sensor nodes. An improved SMDP reinforcement decision algorithm is adopted to adaptively adjust the communication rate and protocol when there is interference or node failure in the wireless communication network, improving the stability and reliability of the network. A node multi-hop communication path and a multi-path transmission mechanism are also adopted to solve the problems in the wireless communication network, avoiding communication interruption and data loss.

[0046] A processing and scheduling module, which performs task scheduling and resource allocation of sensor nodes through an adaptive scheduling strategy model. The strategy model dynamically adjusts the monitoring tasks and data collection frequencies of sensor nodes according to the needs of agricultural production and the monitoring results of the agricultural environment, improving the flexibility and real-time nature of data collection.

[0047] A power management module, which dynamically adjusts the power supply strategy by using an energy management model, manages the energy consumption of sensor nodes through a sleep-wake mechanism, reduces the energy consumption without affecting data collection, and ensures the stable operation of sensor nodes at the same time.

[0048] The real-time monitoring module conducts real-time data monitoring, historical data query, and remote control of environmental parameters and wireless communication status parameters in the monitoring area of facility agriculture through a visualization platform, and uses a user interface to implement the interactive operation between the user and the visualization platform, helping users promptly grasp the status and situation of agricultural production.

[0049] The multi-type low-power sensor chipset includes different types of sensor chips, such as temperature sensor chips, humidity sensor chips, soil humidity sensor chips, light sensor chips, carbon dioxide sensor chips, conductivity sensor chips, etc., to meet the multi-dimensional monitoring requirements of the agricultural production environment. Select corresponding sensor chips according to the characteristics of different crops and production environments. For example, temperature and humidity sensor chips can be used to monitor the temperature and humidity changes in the greenhouse, soil humidity sensor chips can be used to monitor the water content in the soil, light sensor chips can be used to monitor the light conditions required by plants, carbon dioxide sensor chips can be used to monitor the CO2 concentration in the greenhouse, and conductivity sensor chips can be used to monitor the fertility and salinity of the soil, etc.

[0050] Using the multi-type low-power sensor chipset to collect agricultural environment information can achieve all-round monitoring and real-time control of the agricultural production process and environment, and then realize precise management and optimized production of crops.

[0051] In the monitoring of facility agriculture, the improved SMDP (Semi-Markov Decision Process) reinforcement decision algorithm is used to dynamically adjust the communication strategy to adapt to changing factors such as communication distance, data volume, network load, and channel quality. The algorithm realizes the iterative optimization of the communication strategy by defining state parameters, communication strategies, and calculation parameters, and combining the reward and penalty terms of the state utility function, so as to ensure the efficiency and reliability of data transmission.

[0052] As shown in the Figure 2 appendix, the specific implementation steps of the working method of the improved SMDP reinforcement decision algorithm are as follows:

[0053] S1. Parameter initialization;

[0054] Define state parameters, including channel quality, network topology, transmission rate, and protocol type, etc. These parameters jointly describe the current state of the wireless communication network. According to the requirements of facility agriculture monitoring and the performance indicators of the wireless communication network, formulate multiple possible communication strategies. Initialize the calculation parameters, including learning rate, behavior evaluation value function, discount factor, and strategy parameters, etc. These parameters are used for subsequent strategy iteration and optimization.

[0055] S2. Evaluate the advantages and disadvantages of different communication strategies;

[0056] When the data transmission success rate of the wireless communication network is higher than the preset threshold, the time taken for the data to be successfully transmitted to the aggregation node is less than the preset threshold, or the number of sensor nodes simultaneously transmitting data in the current wireless communication network increases, the reward item is increased. When the above conditions are not met, the penalty item is increased. Combining the reward item and the penalty item, a state utility function is designed to evaluate the advantages and disadvantages of different communication strategies.

[0057] S3. Iterative optimization of communication strategies;

[0058] According to the current communication strategy, select appropriate communication protocols and communication rates in the communication protocol library of the wireless communication network. Record the state transition, cumulative reward or penalty, and the state after the transition of the wireless communication network during the execution of the action. Based on the Bellman equation and the recorded data, calculate the action evaluation value. Sort and select the optimal strategy: Sort the expected utility values of the communication strategies and select the communication strategy with the highest expected utility value as the currently selected strategy.

[0059] S4. Execute data transmission and update parameters;

[0060] Execute the data transmission operation according to the currently selected communication strategy. Update the state parameters and calculation parameters according to the transmission result. Record the current state and the corresponding action evaluation value in the historical state list.

[0061] S5. Adjust the value of the action evaluation function and the selection of communication strategies;

[0062] Train according to the historical state list and the action evaluation function. Continuously adjust the value of the action evaluation function and the selection of communication strategies to optimize the communication performance.

[0063] S6. Dynamically adjust communication strategies;

[0064] Continuously monitor changing factors such as communication distance, data volume, network load, and channel quality. Dynamically adjust the communication strategy according to the changing factors, and repeat steps S2 - S6 until the preset end conditions are met (such as reaching the maximum number of iterations, the data transmission success rate stabilizing at a certain level, etc.).

[0065] By implementing the improved SMDP (Semi - Markov Decision Process) reinforcement decision algorithm, the facility agriculture monitoring system can dynamically adjust the communication strategy to adapt to different communication environments and requirements. This can not only improve the efficiency and reliability of data transmission, but also reduce energy consumption and costs, providing a more intelligent and efficient solution for facility agriculture monitoring.

[0066] Compared with the traditional SMDP (Semi - Markov Decision Process) reinforcement decision algorithm, the improvements of the improved SMDP reinforcement decision algorithm are mainly reflected in the following aspects:

[0067] 1. Adaptability of Dynamic State Parameters and Communication Strategies:

[0068] The improved algorithm not only considers basic state parameters such as channel quality and network topology structure, but also specifically introduces specific parameters for the monitoring requirements of facility agriculture, such as data volume, network load, etc., enabling the algorithm to more comprehensively reflect the actual situation of the wireless communication network. The communication strategy is dynamically adjusted according to these dynamically changing state parameters, improving the adaptability and flexibility of the algorithm.

[0069] 2. Optimization of Reward and Punishment Mechanisms:

[0070] The improved algorithm more accurately evaluates the advantages and disadvantages of different communication strategies by adding reward items and punishment items. This includes key indicators such as data transmission success rate, time taken for data to be successfully transmitted to the sink node, and the number of sensor nodes transmitting data simultaneously. This reward and punishment mechanism makes the algorithm pay more attention to the actual application effect when optimizing the communication strategy, thereby improving the reliability and efficiency of data transmission.

[0071] 3. Iterative Optimization and Strategy Selection:

[0072] The improved algorithm adopts an iterative optimization method based on behavior evaluation values. By continuously trying and adjusting communication strategies, and recording state transitions and reward / punishment data during the execution process, it gradually optimizes the communication strategy. The algorithm also sorts the communication strategies according to the expected utility value and selects the optimal strategy as the currently executed strategy, thus ensuring that each data transmission can achieve the best effect.

[0073] 4. Utilization of Historical States and Behavior Evaluation Values:

[0074] The improved algorithm records the current state and the corresponding behavior evaluation values in the historical state list for subsequent training and adjustment. The accumulation and utilization of this historical data enable the algorithm to continuously learn and improve, thus adapting to the continuously changing network environment and monitoring requirements.

[0075] 5. Ability to Dynamically Adjust Communication Strategies:

[0076] The improved algorithm can dynamically adjust communication strategies according to real-time factors such as communication distance, data volume, network load, and channel quality during the communication process. This dynamic adjustment ability enables the algorithm to better handle sudden situations and uncertain factors in the network, thereby ensuring the stability and reliability of data transmission.

[0077] In summary, the improved SMDP reinforcement decision-making algorithm has been improved and enhanced in terms of the adaptability of dynamic state parameters and communication strategies, the optimization of reward and punishment mechanisms, iterative optimization and policy selection, the utilization of historical state and behavior evaluation values, and the ability to dynamically adjust communication strategies. This makes the algorithm have higher practical value and performance in application scenarios such as facility agriculture monitoring.

[0078] The hardware working environment of the improved SMDP reinforcement decision-making algorithm mainly includes the following parts:

[0079] 1. Policy executor: The policy executor is a key component of the SMDP reinforcement decision-making algorithm. It is used to execute the improved decision-making policy and obtain environmental feedback information by interacting with the environment.

[0080] 2. Data storage device: The data storage device is used to store the relevant data of the SMDP reinforcement decision-making algorithm, including decision-making policy data, environmental feedback data, historical data, etc.

[0081] 3. Hardware platform: The hardware platform mainly includes hardware devices such as policy executors and data storage devices, and transmits data through a wireless network.

[0082] 4. Controller: The controller is mainly used to manage and control the execution process of the algorithm. By managing the algorithm and coordinating the cooperation between various hardware devices, smooth operation between devices can be achieved.

[0083] 5. Power supply device: The hardware working environment of the SMDP reinforcement decision-making algorithm requires a stable power supply environment to ensure that each hardware device can work properly. At the same time, energy-saving strategies need to be considered to use less energy to reduce energy costs and environmental impacts.

[0084] 6. Sensor nodes: In practical applications, the SMDP reinforcement decision-making algorithm usually needs to collect environmental data. Therefore, sensor nodes need to be equipped, including temperature sensors, humidity sensors, light sensors, carbon dioxide sensors, etc., so as to obtain environmental feedback information.

[0085] The above hardware working environment is the basic part of the improved SMDP reinforcement decision-making algorithm. Through the cooperation and optimization between these hardware devices, the efficiency and stability of the algorithm system can be maximally improved.

[0086] The laboratory configuration uses a computer with an Intel Core i9 processor, 64G of memory, and 128G of external storage. Simulation software is used to establish a simulation environment. For the on-site experimental environment settings, the simulation data accuracy is 95%, and the algorithm running error does not exceed 2.5%. Comparative experiments are conducted using the improved SMDP reinforcement decision algorithm (Group A) and the SMDP reinforcement decision algorithm (Group B). Five scenarios are manually set, and the improved SMDP reinforcement decision algorithm (Group A) and the SMDP reinforcement decision algorithm (Group B) are used to select the communication rate and protocol respectively. The decision-making time, success rate, and communication efficiency are compared and recorded in Table 1.

[0087]

[0088] From the experimental results, there are slight differences in the performance of the improved SMDP reinforcement decision algorithm and the ordinary SMDP reinforcement decision algorithm in different scenarios. In terms of communication rate, the improved SMDP reinforcement decision algorithm performs better, with higher communication efficiency and success rate. In terms of protocol selection, the difference between the two algorithms is not very obvious, and it needs to be selected according to specific situations. In addition, in terms of decision-making time, the improved SMDP reinforcement decision algorithm is slightly better than the ordinary SMDP reinforcement decision algorithm.

[0089] The communication protocol library includes the Zigbee short-range wireless sensor network protocol, LoRa long-range low-power wireless communication protocol, Wi-Fi wireless local area network protocol, and NB-IoT narrowband Internet of Things protocol. Different communication protocols can be selected according to the actual needs of the wireless communication network to achieve communication between devices.

[0090] The Zigbee protocol is a low-power, low-rate wireless sensor network protocol, suitable for device communication with low data rate and low power requirements. The LoRa protocol is an Internet of Things communication protocol for long-distance communication, which has the characteristics of long-distance communication, low power consumption, and low data rate, and is suitable for occasions such as agricultural applications that require long-distance transmission of information. The Wi-Fi protocol is a high-rate, short-range wireless network protocol, suitable for occasions that require high-rate, short-range communication, such as indoor environments. And NB-IoT is a wireless communication technology based on cellular networks, which has the characteristics of wide coverage, low power consumption, and low cost, and is suitable for the communication of large-scale Internet of Things devices.

[0091] Select an appropriate communication protocol according to the specific requirements of facility agriculture monitoring. For example, for application scenarios with low power consumption, low cost, and short communication distances, the Zigbee protocol can be selected; for application scenarios that require long-distance communication and have low power consumption requirements, the LoRa protocol can be selected; for application scenarios that require high-rate data transmission and wide coverage, the Wi-Fi protocol can be selected; for application scenarios that need to utilize the existing cellular network infrastructure and have wide coverage, the NB-IoT protocol can be selected.

[0092] In some cases, it may be necessary to use multiple protocols in combination to achieve more comprehensive monitoring and communication functions. For example, the Zigbee protocol can be used for short-distance communication between sensor nodes, while the LoRa or NB-IoT protocol can be used for long-distance communication between sensor nodes and remote data centers. At the same time, it is necessary to ensure the interoperability between different protocols to achieve seamless data transmission and processing.

[0093] During the implementation process, it is necessary to optimize and manage the network to ensure the reliability and stability of data transmission. For example, the reliability and stability of the network can be improved by adjusting network parameters, optimizing routing strategies, adding redundant nodes, etc. At the same time, the network needs to be regularly maintained and monitored to detect and solve problems in a timely manner.

[0094] Therefore, the diverse selection of the above communication protocol library can choose different protocols according to the needs of the wireless communication network, thus fully meeting the facility agriculture monitoring needs in different scenarios.

[0095] The multi-hop communication path of nodes realizes cross-node data transmission through data relay transmission between multiple nodes. The specific implementation process of the multi-path transmission mechanism is as follows: The data packet to be transmitted is first divided into multiple data blocks. Then, multiple paths are established between nodes for data transmission. Each path consists of multiple nodes, and data is transmitted between nodes through data relay. The data blocks are transmitted through multiple paths, and data blocks can be transmitted in parallel between different paths, thereby improving the data transmission efficiency. To improve the stability and reliability of the transmission, a standby drive card is used during the transmission process to realize path switching when the transmission path fails.

[0096] Specifically, the standby drive card package includes a standby transmission path. The nodes in the standby path can perform data relay transmission and perform path switching when the main path fails to achieve the continuity of data transmission. At the same time, to improve the transmission success rate, the standby drive card package can also adopt a redundant transmission mechanism. When the main path transmission fails, the standby path can perform multiple retries to ensure the successful transmission of data.

[0097] In summary, by splitting the data to be transmitted into multiple data blocks for multi-path transmission and adopting a spare drive card package to implement path switching and redundant transmission mechanisms in case of transmission path failures, the efficiency, reliability, and stability of data transmission can be effectively improved, and the transmission performance of the facility agriculture monitoring system can be enhanced.

[0098] The spare drive card includes a high-speed serial expansion bus PCIe and a spare control drive circuit. The QOS anti-delay blocking service engine is used to achieve end-to-end communication between the failed transmission path and the spare transmission path. Among them, the high-speed serial expansion bus PCIe is the main data transmission channel, and the spare control drive circuit is connected to the PCIe to realize the data transmission channel of the spare path. The QoS anti-delay blocking service engine is arranged on both the main path and the spare path. During the transmission process, if a failure occurs in the main path, the QoS anti-delay blocking service engine will automatically identify the failure and switch the transmission path to the spare path, and the data transmission will not be affected. The QoS anti-delay blocking service engine supports end-to-end communication, can identify the states of the main path and the spare path, and makes dynamic adjustments during the data transmission process to ensure the smoothness and reliability of data transmission.

[0099] The spare drive card package can achieve main-backup switching, seamlessly switch to the spare path when a failure occurs in the main path, to ensure the continuity of data transmission, and enhance the reliability and stability of the system. And adopting the QoS anti-delay blocking service engine to achieve end-to-end communication can reduce the delay of data transmission and realize automatic failure switching, which can more effectively ensure the normal operation of the facility agriculture monitoring system.

[0100] The adaptive scheduling strategy model includes an input layer, a data layer, a model layer, an algorithm layer, an optimization layer, and an output layer. As shown in the appendix Figure 3 The working method of the adaptive scheduling strategy model includes the following steps:

[0101] Step 1: Receive data;

[0102] Receive the user's requirements through the user interface or API interface, including the monitoring area, monitoring indicators, data collection frequency, etc. Convert the format of the user's requirements to ensure that the data format is compatible with the model. Receive the environmental parameters of the facility agriculture monitoring area through the sensor network, such as temperature, humidity, light intensity, and carbon dioxide concentration, etc. Conduct preliminary verification and format conversion on the environmental parameters to ensure the data quality. Pass the user's requirements and environmental parameters to the data layer for further processing.

[0103] Step 2: Preprocess and integrate data;

[0104] Clean the received data to remove outliers and duplicate data. Standardize the data to ensure that data from different sensor nodes has the same dimension and range. Use the principal component analysis (PCA) method to integrate data from different sensor nodes and extract the main features. Generate a dataset to provide a basis for subsequent model establishment and algorithm application.

[0105] Step 3: Establish an optimal scheduling mathematical model;

[0106] Extract the constraint conditions from the generated dataset, including the calculation scale, objective function, constraint conditions, and variable range, etc. Based on the constraint conditions, establish an optimal scheduling mathematical model, such as a linear programming model, integer programming model, or mixed integer programming model, etc. The model should be able to reflect the relationship between the task scheduling of sensor nodes and the data acquisition frequency, as well as the balance between resource constraints and the objective function.

[0107] Step 4: Classification and prediction;

[0108] Select a suitable time series prediction algorithm, such as ARIMA, LSTM, GRU, etc., to classify and predict the dataset. The algorithm should be able to discover and mine the potential laws and patterns in the data, such as seasonal variations, periodic fluctuations, etc. Dynamically adjust the monitoring tasks and data acquisition frequency of sensor nodes according to the classification and prediction results. The adjustment should be based on real-time data and model prediction results to ensure the accuracy and timeliness of data acquisition.

[0109] Step 5: Iterative calculation and parameter correction;

[0110] Use an iterative algorithm to solve the model, such as a genetic algorithm, particle swarm algorithm, etc. During the iterative process, continuously correct the model parameters to improve the model accuracy and performance. Compare the difference between the calculation result and the true value to correct the model parameters. The correction process should be based on statistical methods and optimization algorithms to ensure the accuracy and effectiveness of parameter correction. Set reasonable thresholds and the number of iterations to ensure that the model can quickly converge to the optimal solution during the iterative process. The selection of the threshold and the number of iterations should be based on the balance between model performance and computing resources. Use parallel computing to distribute the computing tasks to multiple processors or computing nodes. Parallel computing should be able to accelerate the model solving process and improve the computing speed.

[0111] Step 6: Dynamic scheduling and task allocation;

[0112] Output the model training results to the processing and scheduling module, including the optimal scheduling scheme, the task scheduling of sensor nodes, and the data acquisition frequency, etc. Dynamically schedule and allocate tasks to sensor nodes according to the model training results. The scheduling and allocation should be based on real-time data and model prediction results to ensure the accuracy and timeliness of monitoring tasks.

[0113] The time series prediction algorithm predicts the change of environmental parameters at the next moment based on the environmental parameters at historical moments. The data collection dataset of environmental parameters at historical moments is , and the environmental parameter feature dataset that affects the monitoring tasks of sensor nodes is . The environmental parameter feature prediction output function that affects the monitoring tasks of sensor nodes at time t+1 is:

[0114]

[0115] In the formula, is the environmental parameter feature that affects the monitoring tasks of sensor nodes at time t+1, is the environmental parameter feature prediction weighting function, which is used to adjust the influence degree of different historical data points on the predicted value, is the environmental parameter feature that affects the monitoring tasks of sensor nodes at time t, is the environmental parameter feature that affects the monitoring tasks of sensor nodes at time t-1, is the environmental parameter collection data at time t, is the environmental parameter collection data at time t-1, is the ordinal number of the environmental parameter collection moment, .

[0116] The hardware working environment of the adaptive scheduling strategy model mainly includes the following parts:

[0117] 1. Sensor nodes: Sensor nodes are the key components of the adaptive scheduling strategy model. They are used to collect and record environmental parameter data at monitoring points, including meteorological data, soil data, water quality data, etc.

[0118] 2. Data storage devices: Data storage devices are used to store the original monitoring data, preprocessed data, feature data, and adaptive scheduling strategy model collected by sensor nodes.

[0119] 3. Time series prediction model: The adaptive scheduling strategy model can be trained using models such as recurrent neural network (RNN) and long short-term memory network (LSTM), and use historical datasets for prediction.

[0120] 4. Hardware platform: The hardware platform of the adaptive scheduling strategy model mainly includes hardware devices such as sensor nodes, data storage devices, and time series prediction models, and uses wireless networks to transmit the collected data and prediction information.

[0121] 5. Controller: The controller is mainly used to manage and control the collection and prediction processes of sensor nodes, and through management algorithms and coordination of the cooperation between various hardware devices, to achieve smooth operation between devices.

[0122] 6. Power supply equipment: The hardware working environment of the adaptive scheduling strategy model requires a stable power supply environment to ensure that each hardware device can work properly. At the same time, energy-saving strategies need to be considered to use less energy to reduce energy costs and environmental impacts.

[0123] The above hardware working environment is the basic part of the adaptive scheduling strategy model. Through the cooperation and optimization among these hardware devices, the efficiency and stability of the facility agriculture monitoring system can be maximized.

[0124] The laboratory configuration uses a computer with an Intel Core i9 processor, 64G of memory, and 128G of external storage, and uses simulation software to establish a simulation environment. For the on-site experimental environment setting, the simulation data accuracy is 95%, and the algorithm running error does not exceed 2.5%. Comparative experiments are carried out using the adaptive scheduling strategy model (Group A) and the ant colony optimization algorithm (Group B) respectively. Five scenarios are set manually, and the adaptive scheduling strategy model (Group A) and the ant colony optimization algorithm (Group B) are used respectively for task scheduling and resource allocation of sensor nodes. The matching accuracy and time consumption are used to evaluate the effect of the algorithm, which is recorded in Table 2.

[0125]

[0126] According to the analysis of the experimental results in Table 2, we can find that considering both the matching accuracy and time consumption, the adaptive scheduling strategy model (Group A) has better effects in task scheduling and resource allocation of sensor nodes. It should be noted that this conclusion is based on the experimental scenarios and parameters set manually by us. In actual applications, different algorithms need to be selected considering the actual requirements and characteristics of the scenarios.

[0127] The energy management model adopts edge computing. The current working state and power usage of the sensor nodes are compared with the preset values, and the working intensity of the sensor nodes is obtained through comparison. According to the working intensity of the sensor nodes, the corresponding power working mode is matched to use the power in the optimal way to achieve the maximum energy saving. When the sensor nodes collect data, the sleep and wake-up cycles of the sensor nodes are automatically adjusted according to the monitoring requirements and the states of the sensor nodes. The specific implementation methods are as follows:

[0128] 1) Set the working cycle of the sensor nodes. When the sensor nodes are idle, the nodes are put into sleep to reduce energy consumption, and at the same time, the wake-up cycle is set to ensure that the nodes can respond to monitoring requirements in a timely manner.

[0129] 2) When the sensor node continuously monitors and outputs data, measures such as an energy detector or a temperature detector are used to monitor the power and temperature of the node. When the power of the sensor node is too low or the temperature is too high, the node goes into sleep mode to reduce energy consumption and prevent node damage.

[0130] 3) When the sensor node receives a monitoring requirement, a wake-up operation is performed, and the wake-up cycle is adjusted according to the monitoring requirement to ensure that the node can respond to the monitoring requirement in a timely manner.

[0131] By adopting the above energy management model and implementing the sleep-wake mechanism, the energy consumption and maintenance cost of the sensor node can be effectively reduced, thereby improving the stability and reliability of the facility agriculture monitoring system.

[0132] So far, the technical solution of the present invention has been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the protection scope of the present invention.

Claims

1. A facility agriculture monitoring system based on adaptive wireless communication, characterized in that: include: A monitoring sensor module, wherein the monitoring sensor module uses multiple sensor nodes to collect environmental parameters within the facility agriculture monitoring area. The sensor nodes have built-in multiple types of low-power sensor chip groups. The monitoring sensor module cross-validates and compares the data collected by different types of sensor chips through a collaborative mechanism to remove erroneous data; A wireless communication module, which uses a wireless communication network to realize data communication and information exchange between different sensor nodes, and sends and receives data collected by different sensor nodes to a cloud aggregation node for further processing. The wireless communication module uses an improved SMDP enhanced decision algorithm to adaptively adjust the communication rate and communication protocol of the wireless communication network, and establishes a node multi-hop communication path and a multi-path transmission mechanism to deal with communication interference and failures in the wireless communication network; The improved SMDP enhanced decision algorithm dynamically adjusts the communication strategy according to the communication distance, data volume, network load and channel quality. The working method of the improved SMDP enhanced decision algorithm includes the following steps: S1. Define the state parameters and communication strategy of the wireless communication network and initialize the calculation parameters, wherein the state parameters include at least channel quality, network topology, transmission rate and protocol type, and the calculation parameters include learning rate, behavior evaluation value function, discount factor and strategy parameters; S2. According to the needs of facility agriculture monitoring and the performance indicators of the wireless communication network, the reward and penalty items of the state utility function are increased to evaluate the advantages and disadvantages of different communication strategies. When the data transmission success rate of the wireless communication network is less than the preset threshold, the reward item is increased. When the time taken for the data to be successfully transmitted to the sink node is less than the preset threshold, the reward item is increased. When the number of sensor nodes transmitting data simultaneously in the current wireless communication network is greater than the number of sensor nodes at the previous moment, the reward item is increased. Otherwise, the penalty item is increased. S3, iteratively optimize the communication strategy based on the behavior evaluation value, select the communication protocol and communication rate in the communication protocol library of the wireless communication network according to the current communication strategy and execute the action, record the state transition, accumulated rewards or penalties and the state after the transition of the wireless communication network during the execution of the action, and calculate the behavior evaluation value based on the Bellman equation and the recorded data, sort the expected utility values ​​of the communication strategies, and select the communication strategy with the highest expected utility value as the currently selected strategy, wherein the communication protocol library includes at least the Zigbee short-range wireless sensor network protocol, the LoRa long-range low-power wireless communication protocol, the Wi-Fi wireless local area network protocol and the NB-IoT narrowband Internet of Things protocol; S4, performing data transmission operation according to the currently selected communication strategy, and updating the state parameters and calculation parameters according to the transmission result, and recording the current state and the corresponding behavior evaluation value into the historical state list; S5, training is performed according to the historical state list and the behavior evaluation value function, and the value of the behavior evaluation value function and the selection of the communication strategy are continuously adjusted; S6. During the communication process, dynamically adjust the communication strategy according to the communication distance, data volume, network load and channel quality factors, and repeat steps S2-S6 until the preset end condition is reached; A processing and scheduling module, wherein the processing and scheduling module uses an adaptive scheduling strategy model to schedule monitoring tasks of multiple sensor nodes, and the adaptive scheduling strategy model dynamically adjusts the monitoring tasks and data collection frequency of the sensor nodes according to the needs of agricultural production and the monitoring results of facility agricultural environmental parameters; The adaptive scheduling strategy model includes an input layer, a data layer, a model layer, an algorithm layer, an optimization layer and an output layer. The working method of the adaptive scheduling strategy model includes the following steps: Step 1: Receive user requirements and environmental parameters of the facility agriculture monitoring area through the input layer, convert the format, and then pass the data to the data layer for further processing. The environmental parameters include at least temperature, humidity, light intensity and carbon dioxide concentration; Step 2: The data layer preprocesses and cleans the data transmitted by the input layer, and integrates the data from different sensor nodes using principal component analysis to generate a data set; Step 3, the model layer obtains restriction conditions from the generated data set and establishes an optimal scheduling mathematical model based on the restriction conditions, wherein the restriction conditions include calculation scale, objective function, constraint conditions and variable range; Step 4: The algorithm layer classifies and predicts the data set based on the time series prediction algorithm to discover and mine the potential laws and patterns in the data, and dynamically adjusts the monitoring tasks and data collection frequency of the sensor nodes according to the classification and prediction results; The time series prediction algorithm predicts the change of environmental parameters at the next moment based on the environmental parameters at the historical moment. The environmental parameter collection data set at the historical moment is Y={y1,y2,...,y t-1 ,y t }, the characteristic data set of environmental parameters that affect the monitoring task of sensor nodes is X = {x1, x2, ..., x t-1 ,x t }, the prediction output function of the environmental parameter characteristics that affect the monitoring task of the sensor node at time t+1 is: In the formula, x t+1 is the environmental parameter characteristics that affect the monitoring task of the sensor node at time t+1, is the weighted function for predicting the environmental parameter characteristics that affect the monitoring task of the sensor node, which is used to adjust the influence of different historical data points on the predicted value. t is the environmental parameter characteristic that affects the monitoring task of the sensor node at time t, x t-1 is the environmental parameter characteristic that affects the monitoring task of the sensor node at time t-1, y t Collect data for environmental parameters at time t, y t-1 Collect data of environmental parameters at time t-1, where t is the ordinal number of the time when the environmental parameters are collected, and t≥1; Step 5: Perform iterative calculation, parameter correction, and comparison between calculation results and true values ​​through the optimization layer, and set the threshold and number of iterations through adaptive parameter selection. The optimization layer uses parallel computing to distribute the calculation tasks to multiple processors or computing nodes to increase the calculation speed. Step 6: The output layer transmits the model training results, the task scheduling of the nodes, and the data collection frequency to the processing scheduling module to realize dynamic scheduling and task allocation of the sensor nodes; A power management module, which uses an energy management model to dynamically adjust the power supply strategy. The energy management model manages the energy consumption of sensor nodes through a sleep and wake-up mechanism, and dynamically adjusts the power output according to the energy consumption requirements of the sensor nodes, wireless communication modules, and processing and scheduling modules; A real-time monitoring module performs real-time data monitoring, historical data query and remote control of environmental parameters and wireless communication status parameters in the facility agriculture monitoring area through a visualization platform, and uses a user interface to realize interactive operations between the user and the visualization platform.

2. The facility agriculture monitoring system based on adaptive wireless communication according to claim 1, characterized in that: The multi-type low-power consumption sensor chipset comprises at least a temperature sensor chip, a humidity sensor chip, a soil humidity sensor chip, a light sensor chip, a carbon dioxide sensor chip and a conductivity sensor chip.

3. The facility agriculture monitoring system based on adaptive wireless communication according to claim 1 is characterized in that: The node multi-hop communication path realizes cross-node data transmission through data relay transmission between multiple nodes. The multi-path transmission mechanism divides the data packet to be transmitted into multiple data blocks, transmits them through different transmission paths, and uses a backup driver card to realize path switching when the transmission path fails.

4. The facility agriculture monitoring system based on adaptive wireless communication according to claim 3 is characterized in that: The standby drive card comprises a high-speed serial expansion bus PCIe and a standby control drive circuit. The high-speed serial expansion bus PCIe adopts a QOS anti-delay blocking service engine to realize end-to-end communication between a faulty transmission path and a standby transmission path.

5. The facility agriculture monitoring system based on adaptive wireless communication according to claim 1, characterized in that: The energy management model uses edge computing to compare the current working status and power usage of the sensor node with the preset value to obtain the working intensity of the sensor node, and matches the power working mode based on the working intensity. The sleep and wake-up mechanism automatically adjusts the sleep and wake-up cycle of the sensor node according to the monitoring requirements and the status of the sensor node.

Images