Design And Implementation Of Multi-Sensor Water Quality Monitoring System Based On LoRa

China,with its advantageous geographical location,has over 18,000 kilometers of coastline,more than 26.67 million hectares of inland water areas,and over 1,000 economically valuable aquatic products,making it a major country in aquaculture worldwide.In recent years,the worsening water pollution caused by human activities has led to a significant reduction in the production of large seaweeds,fish,shellfish,and other aquatic products,with some species even nearing extinction.In this context,modern aquaculture is inevitably moving towards industrialization and intelligentization.Smart water quality monitoring allows real-time and online acquisition of key environmental factors in aquatic product farming,thereby improving productivity and quality,achieving both economic benefits and environmental sustainability,and becoming one of the important technologies in the current industry.In the current industrialized aquaculture,wireless communication methods such as Zigbee,Wi Fi,and GPRS are mostly used for water quality monitoring data transmission.However,these technologies suffer from short communication distances,poor interference resistance,and high power consumption,making it difficult to meet the demands of large-scale aquaculture.This thesis designs a multi-parameter water quality monitoring system based on LoRa technology.Multi-sensors and the main controller are used to build a collection and integration system to realize data acquisition,analysis and storage.LoRa technology with low power consumption and long distance communication advantages was used for wireless transmission of sensor data collection.Data analysis and neural network combined with optimization algorithm are used to construct a prediction model to realize noise reduction and prediction of key parameters of water quality,and ensure long-term controllability of the breeding environment.The main research contents of the intelligent water quality monitoring system are as follows:(1)A comparison and analysis of several commonly used wireless communication technologies were conducted in accordance with specific application scenarios.The application of LoRa communication technology in the aquaculture water quality monitoring system was implemented,and the overall system architecture was determined.(2)The hardware design of the system was carried out,including the acquisition nodes and the aggregation node.The acquisition nodes collect data such as temperature,dissolved oxygen saturation,turbidity,and p H values in the aquaculture area through sensors.The data is then transmitted upward to the aggregation node using LoRa technology,and subsequently uploaded to the cloud server by the aggregation node via a 4G communication network.(3)The STM32L475 was used as the main controller,and the RT-Thread embedded real-time operating system was ported to achieve scheduled collection,packaging,and uploading of various water quality data,as well as storing the data in an SD card.A UV lamp module was controlled to prevent biological attachment and extend the sensor’s lifespan.(4)The monitoring interface was designed in conjunction with the Ali Cloud Io T platform.This interface can receive uploaded data and provide real-time monitoring,ultimately displaying the data in the form of line charts and dashboards.Thresholds for water quality data were set,and users are promptly notified when the water quality parameters exceed the thresholds for appropriate action.A water quality parameter database was established to facilitate user operations on historical data.(5)A prediction method was designed that combines the Wavelet Transform(WT)algorithm and the Long Short-Term Memory(LSTM)neural network.This method enables the decomposition and reconstruction of data,effectively eliminating data noise.The WT-LSTM model was then constructed using the Long Short-Term Memory neural network,achieving the prediction of water quality parameters.(6)Hardware testing for the acquisition nodes,LoRa communication distance testing,functionality testing for data visualization,and comparative analysis of prediction models were conducted.The acquisition nodes are capable of real-time data collection,uploading,and receiving instructions from the host computer.The LoRa communication distance can meet the requirements within a communication range of 2 km with a packet loss rate of no more than 10% and good communication quality.The data visualization interface dynamically displays real-time water quality data,including line graphs and dashboards on both the web and mobile platforms,while also monitoring device status.Comparative analysis of the prediction models shows that the WT-LSTM model can effectively reduce the data noise issues present in the LSTM and RNN models’ prediction results,with an RMSE reduction of31.38%,MAE reduction of 31.77%,and MAPE reduction of 33.33% compared to the WT-RNN model.Thus,the prediction accuracy of dissolved oxygen saturation is better improved.