DDGS feed detection method and system based on wireless transmission

Abstract

The application discloses a DDGS feed detection method and system based on wireless transmission, relates to the technical field of intelligent internet-of-things monitoring devices, and aims to solve the problems of incomplete data collection, unstable communication link, large sensor data error and inaccurate abnormal positioning in the existing DDGS yard monitoring; the application proposes to predict channel attenuation through RSSI and bit error rate data, combine adaptive frequency hopping and MIMO antenna to optimize signal transmission, utilize a blind source separation algorithm to suppress multipath effect and improve data accuracy, and combine inertial navigation and UWB ranging to realize high-precision positioning and generate a heat map to mark an abnormal area; meanwhile, an edge gateway adopts a BP neural network to correct sensor error and improve measurement reliability; the application significantly improves monitoring accuracy, stability and real-time performance and improves the safety management level of the DDGS yard.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent Internet of Things monitoring devices, and more particularly to a DDGS feed detection method and system based on wireless transmission. BACKGROUND

[0002] With the rapid development of information technology and wireless communication, emerging technologies such as the Internet of Things and big data are penetrating various industries. The agricultural and feed production fields have ushered in an era of intelligent monitoring, promoting the popularization of unmanned early warning systems, providing strong technical support for product quality and warehouse safety management, and laying the foundation for industry upgrading.

[0003] Currently, domestic DDGS feed surveys generally use pile storage. Due to the strong thermal conductivity and hydrophilicity of DDGS feed, the pile is prone to insect infestation during storage, and the high bulk density can easily cause mold growth, spontaneous combustion, and other safety hazards. To reduce risks, production and use companies often take measures such as manual mouse control and regular warehouse turnover. However, conventional cable-type temperature and humidity sensors have construction difficulties when laid in the yard, and non-contact detectors cannot accurately reflect the internal environmental parameters of the pile. Therefore, on-site detection mainly relies on manual inspection with a 2-meter temperature probe, which is labor-intensive and has limited coverage, resulting in incomplete data collection and delayed response. In addition, due to the presence of metal equipment in the DDGS yard, uneven distribution of the pile itself, and factors such as humidity and dust, the signal is easily severely attenuated during transmission, leading to unstable communication links. At the same time, when using distributed collection points, the phase interference caused by multipath propagation causes significant deviations in sensor sampling data, making it impossible to accurately reflect temperature, humidity, and PH3 concentration. Furthermore, GPS is easily obstructed inside the yard, and temperature, humidity, and PH3 sensors are easily disturbed in harsh environments, affecting accurate positioning and judgment of abnormal conditions. SUMMARY

[0004] To address the deficiencies of the prior art, the present application discloses a DDGS feed detection method and system based on wireless transmission, aiming to solve the problems raised in the background art.

[0005] The present application adopts the following technical solutions:

[0006] A DDGS feed detection method based on wireless transmission, comprising the following steps:

[0007] Step 1: Based on real-time collected RSSI and bit error rate data, the detection node predicts the channel attenuation trend through a long short-term memory network. If the predicted signal quality is lower than a preset threshold, the wireless operating frequency band is switched through an adaptive frequency hopping mechanism, and the real-time signal is weighted and synthesized through a MIMO antenna array.

[0008] Step 2, based on the optimized wireless channel data in step 1, the probe node extracts the main path signal through a blind source separation algorithm, and calculates the phase offset of different path signals combined with the probe node position topology data to perform phase compensation on the received signal;

[0009] Step 3, when the GPS signal strength is monitored to be lower than the preset threshold, the probe node generates the absolute position of the node based on the inertial navigation module and the UWB ranging data of the adjacent node using a federated filtering algorithm, and outputs the temperature and humidity and PH3 concentration distribution thermodynamic diagram;

[0010] Step 4, the probe node samples data through the PH3 sensor at a preset sampling period, and transmits the collected data and the position information output in step 3 to the edge gateway through the wireless channel, and the edge gateway uses a BP neural network to construct an error correction model, compensates the PH3 sensor data in real time according to the dust concentration and humidity data in the current environment, and outputs the calibrated PH3 data;

[0011] Step 5, if the PH3 concentration of the same probe node is continuously sampled for 3 times and exceeds the preset threshold, and the temperature and humidity meet the abnormal fermentation characteristics, the concentration diffusion path is traced back through the gradient descent algorithm, the abnormal core area is marked combined with the position data output in step 3, and the abnormal data is output to the early warning control host;

[0012] Step 6, the early warning control host uploads the local collected data to the cloud, and the cloud aggregates the data of multiple stockpiles through a federated learning framework to train an abnormal diagnosis model, and feeds back the updated model to the probe node;

[0013] Step 7, collect the battery power and current communication load data of each probe node, and allocate the transmission power through a game theory power allocation model, if the node power is lower than 20% and is not in an abnormal area, switch to LoRa communication mode;

[0014] Step 8, periodically detect the health status of the probe node, including battery power, sensor data drift and wireless communication state; if any data exceeds the preset abnormal threshold, mark it as a fault node and start the adjacent node topology reconstruction, and report the maintenance instruction synchronously.

[0015] Preferably, the working steps of step 2 include:

[0016] Step S1, based on the optimized wireless channel data in step 1, the probe node receives and stores the wireless signal containing multipath effect to form an observation signal matrix , ; wherein represents the number of receiving antennas; represents the number of sampling points;

[0017] Step S2, performing pretreatment on the observation signal matrix , constructing a cost function based on a generalized information maximization method , realizing statistical independence of each independent component, and a formula expression of the cost function is:

[0018] (1)

[0019] In the formula (1), is a demixing matrix, is a signal matrix after pre-whitening processing, used for reducing signal correlation; is an objective function, used for enhancing signal sparsity control; is a regularization coefficient, used for preventing numerical overflow; is an eigenvector matrix of the received observation signal matrix ;

[0020] Step S3, based on the cost function , a demixing matrix is iteratively optimized by a gradient descent method, an optimal independent component matrix is obtained, and a main path signal is selected from the independent component matrix ;

[0021] Step S4, based on three-dimensional coordinates of the detector , , a multi-path channel propagation model is constructed, and a propagation distance of a signal propagating to a scattering point , from a path is:

[0022] (2)

[0023] In the formula (2), , is a spatial coordinate of a reflection point of the path ; , is a signal emission source coordinate; is a path correction factor, used for considering a signal diffraction effect; is an incident angle of the path ;

[0024] Step S5, a phase offset of each path signal is calculated by using the propagation model , and a calculation formula is:

[0025] (3)

[0026] In the formula (3), representative signal wavelength, is a reference position coordinate; is a nonlinear phase compensation weight, used to correct the interference of different path signals;

[0027] Step S6, considering the attenuation characteristics of each path signal, the generalized optimal weight method is used to calculate the comprehensive phase , the calculation formula is:

[0028] (4)

[0029] In formula (4), is a path loss weight; is a path loss exponent correction factor, wherein is a loss adjustment coefficient; is the number of received paths; wherein the path loss weight , the path loss value The calculation formula is:

[0030] (5)

[0031] In formula (5), , is a path attenuation exponent; is a correction factor;

[0032] Step S7, in the real-time processing process, the detection node compensates the received signal according to the phase offset to obtain the compensated signal , and the formula expression of the compensation operation is:

[0033] (6)

[0034] In formula (5), is a phase compensation factor, is a long-distance channel correction coefficient, wherein is a channel nonlinear correction parameter.

[0035] A DDGS feed detection system based on wireless transmission, comprising: an early warning control host and a distributed wireless detection and mouse integrated machine;

[0036] The early warning control host computer comprises an upper computer system, a first wireless communication module, a channel assessment and frequency hopping module and a remote management and control interface; the distributed wireless detection and mouse driving integrated machine comprises a second wireless communication module, an inertial navigation module, a temperature and humidity detection module, a PH3 sensing module, a dust sensing module, an ultrasonic wave mouse driving module, a MIMO antenna array and a lithium battery module; output ends of the inertial navigation module, the temperature and humidity detection module, the PH3 sensing module and the dust sensing module are connected to the upper computer system of the early warning control host computer through the second wireless communication module; an output end of the upper computer system is connected to the inertial navigation module, the temperature and humidity detection module, the PH3 sensing module, the dust sensing module, the ultrasonic wave mouse driving module, the MIMO antenna array and the lithium battery module of the distributed wireless detection and mouse driving integrated machine through the first wireless communication module.

[0037] Based on the above technical solution, the positive beneficial effects of the present application are:

[0038] In view of the problem that artificial inspection is highly dependent and the detection coverage is limited, the present application realizes accurate identification and real-time monitoring of abnormal areas by deploying distributed wireless detection nodes and combining gradient descent algorithm to track concentration diffusion path, avoiding the abnormal lag discovery caused by too long artificial sampling interval in traditional methods, and improving the inspection efficiency and accuracy.

[0039] In view of the problem that the signal inside the DDGS yard is severely attenuated, the present application uses LSTM network to predict the channel attenuation trend, combines adaptive frequency hopping mechanism to optimize the wireless communication link, and simultaneously enhances the signal reception quality through MIMO antenna array, thereby reducing the interference of the stacking environment on the wireless signal, improving the communication stability, and solving the problem of unstable data transmission and high packet loss rate in the existing method.

[0040] In view of the problem of phase interference caused by multipath propagation, the present application uses blind source separation algorithm to extract the main path signal, and combines phase compensation technology to correct the phase offset of different path signals, so as to improve the accuracy of the received signal, thereby reducing the measurement error of the sensor caused by signal interference, and making the temperature and humidity and PH3 concentration data more reliable.

[0041] In view of the problem of low positioning accuracy of GPS signal affected by shielding, the present application uses inertial navigation module and UWB ranging fusion algorithm, and combines federal filtering algorithm to optimize node position data, so that the detection node can still be accurately positioned even in the environment without GPS signal, thereby improving the positioning accuracy of the abnormal area and avoiding misjudgment.

[0042] To address the issue that temperature, humidity, and pH3 sensors are susceptible to interference in harsh environments, this invention employs a BP neural network error correction model. This model uses dust concentration and humidity data to compensate for sensor measurements in real time, improving data accuracy, avoiding monitoring errors caused by environmental changes, and enhancing the ability to detect abnormal situations. Attached Figure Description

[0043] Figure 1 This is a schematic diagram illustrating the steps of a DDGS feed detection method based on wireless transmission according to the present invention.

[0044] Figure 2 This is a framework diagram of the DDGS feed detection system based on wireless transmission according to the present invention;

[0045] Figure 3 This is a schematic diagram of the principle framework of step S1 of the present invention;

[0046] Figure 4 This is a flowchart illustrating the working method of step S5 of the present invention.

[0047] Figure 5 This is a flowchart illustrating the working method of step S6 of the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] In an embodiment, such as Figure 1 As shown: The steps of a DDGS feed detection method based on wireless transmission are as follows: Step 1, based on real-time acquired RSSI and bit error rate data, the detection node predicts the channel attenuation trend through a long short-time memory network. If the predicted signal quality is lower than a preset threshold, the wireless operating frequency band is switched through an adaptive frequency hopping mechanism, and the real-time signal is weighted and synthesized through a MIMO antenna array; Figure 3As shown, the specific working method is: based on the real-time collected RSSI and bit error rate data, the detection node first inputs the collected channel state data as a feature vector into a long short-term memory network, constructs a channel attenuation model in the time series prediction process, and predicts the future signal quality; in the model training and prediction process, the weight is optimized through the historical data to obtain the predicted signal attenuation trend curve; if the prediction result shows that the signal quality is lower than the preset threshold, an adaptive frequency hopping mechanism is used, the energy detection and interference identification algorithm is used to analyze and select the working frequency band in real time, and the software defined radio interface is used for spectrum switching; based on the received multi-channel data after switching, the maximum ratio synthesis method is used to weight and synthesize and spatially filter the signals of each channel through the MIMO antenna array, and the synthesized signal is output as the optimal channel data.

[0050] Step 2, based on the wireless channel data optimized in step 1, the detection node extracts the main path signal through a blind source separation algorithm, and calculates the phase offset of different path signals combined with the detector position topology data to compensate the phase of the received signal; the specific working steps include:

[0051] Step S1, based on the wireless channel data optimized in step 1, the detection node receives and stores the wireless signals containing multipath effects to form an observation signal matrix , ; wherein represents the number of receiving antennas; represents the number of sampling points;

[0052] Step S2, the observation signal matrix is preprocessed through a blind source separation algorithm, a cost function is constructed based on the generalized information maximization method to realize the statistical independence of each independent component, and the formula expression of the cost function is:

[0053] (1)

[0054] In formula (1), is a demixing matrix, is a pre-whitened signal matrix used to reduce signal correlation; is an objective function used to enhance signal sparsity control; is a regularization coefficient used to prevent numerical overflow; is a feature vector matrix of the received observation signal matrix ;

[0055] Step S3, based on the cost function , the demixing matrix is iteratively optimized by the gradient descent method to obtain the optimal independent component matrix and the main path signal is selected from the main path signals ;

[0056] Step S4, constructing a three-dimensional coordinate system based on the detector , , constructing a multi-path channel propagation model, defining the propagation distance of the signal to the scattering point , , of the path is:

[0057] (2)

[0058] In formula (2), , is the spatial coordinate of the path reflection point; , is the signal emission source coordinate; is the path correction factor, which is used to consider the signal diffraction effect; is the incidence angle of the path ;

[0059] Step S5, calculating the phase shift of each path signal using the propagation model , the calculation formula is:

[0060] (3)

[0061] In formula (3), represents the signal wavelength, is the reference position coordinate; is the nonlinear phase compensation weight, which is used to correct the interference of different path signals;

[0062] Step S6, considering the attenuation characteristics of each path signal, and calculating the comprehensive phase using the generalized optimal weight method , the calculation formula is:

[0063] (4)

[0064] In formula (4), is the path loss weight; is the path loss exponential correction factor, wherein is the loss adjustment coefficient; is the number of received paths; wherein the path loss weight , the calculation formula of the path loss value is:

[0065] (5)

[0066] In formula (5), , is a path loss exponent; is a correction factor;

[0067] Step S7, in the real-time processing process, the detection node compensates the received signal according to the phase offset to obtain a compensated signal , and the formula expression of the compensation operation is:

[0068] (6)

[0069] In formula (5), is a phase compensation factor, is a remote channel correction coefficient, wherein is a channel nonlinear correction parameter.

[0070] Step 3, when the GPS signal strength is lower than the preset threshold, the detection node generates the absolute position of the node based on the inertial navigation module and the UWB ranging data of the adjacent node by using the federated filtering algorithm, and outputs the temperature and humidity and PH3 concentration distribution thermodynamic diagram; wherein, the detection node first collects acceleration data , angular velocity data output by the inertial navigation module, and relative distance data obtained by the UWB ranging module of the adjacent node; a state vector is constructed by using the federated filtering algorithm, wherein represents the node position, represents the speed, represents the number of sampling points; and the state transition matrix , the control input matrix and the process noise covariance are set, the prior state estimation of each node is obtained by using the prediction formula and the prediction error covariance, and each local node is predicted; the prediction formula is:

[0071] (7)

[0072] The prediction error covariance is:

[0073] (8)

[0074] In formula (7) and formula (8), represents the acceleration input obtained by the IMU, represents the posterior state estimation at the last moment; represents the process noise; ​​error covariance matrix; then, in the local update stage, the observation vector is obtained by combining the inertial navigation module and the UWB ranging module , define the observation matrix and the measurement noise covariance , calculate the Kalman gain and update the local posterior state estimation and error covariance matrix, the formula for calculating the Kalman gain is:

[0075] (9)

[0076] In formula (9), is the observation matrix, which is used to map the state vector to the measurement space; is the covariance; is the measurement noise; for the nodes participating in federated filtering, the global state estimation is obtained by using the weighted optimal fusion method based on the state estimation and the corresponding covariance of each local node , the formula is:

[0077] (10)

[0078] In formula (10), is the local posterior state estimation of the node; is the corresponding error covariance; is the total number of received paths; finally, the first three-dimensional components of the global state estimation are extracted as the absolute position of the node, and the corresponding temperature, humidity and PH3 concentration data of each node are associated with the absolute position of the node by the Kriging interpolation method to generate the DDGS yard environmental parameter distribution thermodynamic map.

[0079] Step 4: The detection node samples data from the PH3 sensor according to a preset sampling period, and transmits the collected data and the location information output in Step 3 to the edge gateway via a wireless channel. The edge gateway uses a BP neural network to construct an error correction model, which compensates the PH3 sensor data in real time based on the dust concentration and humidity data in the current environment, and outputs calibrated PH3 data. The error correction model includes an input layer, a feature normalization layer, a linear transformation hidden layer, a nonlinear activation layer, a fully connected fusion layer, an output correction layer, and a dynamic update layer. The input layer is used to standardize the raw data from the PH3 sensor and the dust concentration and humidity data collected from environmental detection through digital filtering and normalization algorithms, generating a unified feature vector, and outputting it to the feature normalization layer. The feature normalization layer is used to normalize each dimension using a batch normalization mechanism. The data distribution is adjusted; the linear transformation hidden layer is used to linearly extract sensor drift data using a weight matrix and bias vector; the nonlinear activation layer is used to perform nonlinear mapping using the ReLU activation function to capture the nonlinear interference of dust and humidity on PH3 data; the fully connected fusion layer is used to fuse various hidden features through the weighted superposition mechanism of a multilayer perceptron and output comprehensive correction parameters, which are used to quantitatively reflect the global mapping of environmental factors on PH3 measurement error; the output correction layer is used to perform real-time dynamic compensation on the original PH3 data based on the comprehensive correction parameters using element-wise multiplication to obtain calibrated PH3 concentration data; the dynamic update layer is used to continuously iteratively adjust weights and biases using backpropagation and Adam optimization strategies when there is a deviation between the compensated data and the expected target.

[0080] Step 5: If the PH3 concentration at the same detection node exceeds the preset threshold for three consecutive samplings and the temperature and humidity match the characteristics of abnormal fermentation, the concentration diffusion path is traced in reverse using the gradient descent algorithm. The abnormal core area is marked using the location data output in Step 3, and the abnormal data is output to the early warning control host. Wherein, if... Figure 4 As shown, the edge gateway first constructs a concentration distribution function based on the calibrated PH3 data. ,in Represents a spatial location vector. The value represents the position. The pH3 concentration at the location; then, the edge gateway uses a gradient descent algorithm to analyze the concentration distribution function. The derivative processing is performed to obtain the concentration change rate to determine the local concentration gradient direction, and the initial sampling position is taken as a starting point to start the iterative update. In each iteration, based on the concentration data change rate, the system determines the next search direction by calculating the environmental gradient, and then judges whether the local concentration peak value meets the convergence condition according to the convergence criterion; if yes, the local abnormal area coordinates obtained by the reverse tracking are obtained; if not, the iterative update is continued; then, the edge gateway performs data fusion on the local abnormal area coordinates and the position data output in step 3 through the spatial alignment mechanism and the position data correction algorithm; in the data fusion process, the position data of each node is statistically processed by the weighted average method.

[0081] Step 6, the early warning control host uploads the local collection data to the cloud, and the cloud aggregates the data of multiple yards to train an anomaly diagnosis model through a federated learning framework, and feeds back the updated model to the detection nodes; the specific working method is as follows: after the early warning control host uploads the local data collected by each distributed detection node to the cloud server through a secure transmission protocol, the cloud starts a federated learning framework, and aggregates the local model updates from multiple DDGS yards asynchronously through a secure multi-party computing mechanism; during the aggregation process, each detection node first calculates the local loss function gradient locally using a privacy protection mechanism, generates local model weight update data through a local gradient descent algorithm, and transmits the data to the cloud; after the cloud receives multiple local gradients, it fuses the local update data using a weighted average strategy, and filters and denoises the gradient data using Bayesian inference and differential privacy algorithms; then, the cloud iteratively updates the global anomaly diagnosis model parameters through a stochastic gradient descent method, and feeds back the updated model parameters to each detection node; finally, the detection node matches the updated global parameters of the model with the local model through a local model fusion mechanism, and adjusts the local anomaly detection strategy.

[0082] Step 7, collect the battery power and current communication load data of each detection node, and allocate the transmission power through a game theory power allocation model; if the node power is less than 20% and is not an abnormal area, switch to LoRa communication mode; wherein the game theory power allocation model takes each node as a game participant, constructs a transmission quality and energy consumption utility function, and solves the Nash equilibrium through distributed iteration to obtain the optimal transmission power allocation.

[0083] Step 8, periodically detecting the health status of the detection node, including battery power, sensor data drift and wireless communication state; if any data exceeds the preset abnormal threshold, it is marked as a fault node and the adjacent node topology reconstruction is started, and the maintenance instruction is reported synchronously. Further, when any data is detected to exceed the preset abnormal threshold, the node state is updated to a fault node, based on the node marking, the warning control host sends a state query instruction to the adjacent nodes through the broadcast protocol, collects the latest position information, communication quality parameters and network connection state of each node, then the warning control host calculates the best route according to the node connection weight by the shortest path method, and triggers dynamic route update in the network; the node connection weight is composed of node residual energy, signal quality and link delay.

[0084] As shown in Figure 2 A DDGS feed detection system based on wireless transmission, including: early warning control host and distributed wireless detection and mouse integrated machine; the distributed wireless detection and mouse integrated machine includes n wireless detection and mouse integrated machines; the early warning control host includes host system, first wireless communication module, channel assessment and frequency hopping module and remote management and control interface; the distributed wireless detection and mouse integrated machine includes second wireless communication module, inertial navigation module, temperature and humidity detection module, PH3 sensing module, dust sensing module, ultrasonic mouse driving module, MIMO antenna array and lithium battery module; the output ends of the inertial navigation module, temperature and humidity detection module, PH3 sensing module and dust sensing module are connected to the host system of the early warning control host through the second wireless communication module; the output end of the host system is connected to the inertial navigation module, temperature and humidity detection module, PH3 sensing module, dust sensing module, ultrasonic mouse driving module, MIMO antenna array and lithium battery module of the distributed wireless detection and mouse integrated machine through the first wireless communication module.

[0085] In step 1 of the above embodiment, the long short-term memory network, as a special kind of recurrent neural network (RNN), solves the problems of gradient vanishing and gradient explosion that traditional RNNs face when processing long sequence data. LSTM contains a memory cell (cell state) and three gating structures, namely input gate, forget gate and output gate. In the wireless transmission-based DDGS feed detection method, the detection node inputs the real-time collected RSSI (received signal strength indication) and bit error rate data as feature vectors into LSTM. The input gate decides which new information will be added to the memory cell, the forget gate controls which information in the memory cell will be retained or discarded, and the output gate decides which information in the memory cell will be output for prediction. Through learning and training of historical data, LSTM continuously adjusts the internal weight parameters to build a model that can accurately reflect the channel attenuation trend. In the prediction process, the memory cell will retain important information related to channel attenuation according to the input feature vector and the control of the gating mechanism, thereby realizing the prediction of future signal quality and generating a predicted signal attenuation trend curve.

[0086] When the predicted signal quality by LSTM is lower than the preset threshold, the adaptive frequency hopping mechanism is started. This mechanism first uses the energy detection algorithm to detect the signal energy in the current frequency band to determine whether there is interference in the frequency band and the strength of the interference. Energy detection determines the occupation of the frequency band by analyzing the power spectral density of the received signal. At the same time, the interference identification algorithm further classifies the detected interference to determine the type (such as narrowband interference, wideband interference, etc.) and source of the interference. Based on the results of energy detection and interference identification, the system selects a suitable working frequency band from the pre-set frequency table. Then, the software-defined radio (SDR) interface is used to realize spectrum switching. Software-defined radio defines and controls various functions of the radio through software, enabling the system to flexibly switch between different frequency bands to avoid interference frequency bands and select frequency bands with better signal quality for communication.

[0087] MIMO (Multiple Input Multiple Output) antenna array technology uses multiple transmitting antennas and multiple receiving antennas to improve the performance of the communication system through spatial multiplexing and spatial diversity. In this method, when the probe node switches to a new frequency band, it receives multi-channel data through the MIMO antenna array. The Maximum Ratio Combining (MRC) method is an important technology in MIMO signal processing. MRC assigns weights to each received channel signal based on its signal-to-noise ratio (SNR). The higher the signal-to-noise ratio of a channel, the greater the weight assigned to it. By weighting and combining the signals of each channel according to the weight, and performing spatial filtering processing, the signal-to-noise ratio of the combined signal can be effectively improved, and the strength and quality of the signal can be enhanced, thereby obtaining the optimal channel data output.

[0088] In the actual implementation of step 1, the main hardware includes a signal acquisition module, a processing module, a storage module, a wireless communication module, and a MIMO antenna array. The signal acquisition module is used to collect RSSI and bit error rate data in real time, and can use an integrated wireless signal sensor. The processing module is the core of the probe node, responsible for running the long short-term memory network, the adaptive frequency hopping algorithm, and the MIMO signal processing algorithm, and can use a high-performance microprocessor or embedded processor. The storage module is used to store historical acquisition data, trained LSTM model parameters, and preset frequency table information, and can use flash memory or other non-volatile memory. The wireless communication module is responsible for implementing the transmission and reception of wireless signals, and can use a wireless communication chip that supports multiple frequency bands. The MIMO antenna array is composed of multiple antennas and is used to receive multi-channel signals. Among them, the signal acquisition module is connected to the wireless communication module, and the collected RSSI and bit error rate data are transmitted to the wireless communication module. The wireless communication module transmits the data to the processing module for processing. The processing module is connected to the storage module, reads historical data and model parameters for calculation, and stores the processing results back to the storage module. The processing module also controls the spectrum switching of the wireless communication module and the signal processing operation of the MIMO antenna array. The MIMO antenna array is connected to the wireless communication module, and transmits the received multi-channel signals to the wireless communication module for subsequent processing.

[0089] In actual work, the detection node first collects RSSI and bit error rate data in real time through the signal acquisition module. The collected data is transmitted to the processing module, which inputs it as a feature vector into the trained long short-term memory network to predict the channel attenuation trend. If the prediction result shows that the signal quality is below the preset threshold, the processing module will start the adaptive frequency hopping mechanism, select the appropriate working frequency band through energy detection and interference identification algorithm, and control the wireless communication module to switch the frequency spectrum using the software-defined radio interface. After switching the frequency band, the MIMO antenna array receives multi-channel signals, and the processing module uses the maximum ratio combining method to weight and combine the signals of each channel and perform spatial filtering to obtain the optimal channel data for subsequent DDGS feed detection data transmission.

[0090] Compared with the prior art, through the accurate prediction of the channel attenuation trend by the long short-term memory network, the risk of signal quality decline can be discovered in advance, and adaptive frequency hopping measures can be taken in time to effectively avoid interruptions and instability problems in the signal transmission process, ensuring the reliable transmission of DDGS feed detection data. In addition, the adaptive frequency hopping mechanism can detect and identify interference in real time and quickly switch to a frequency band with little or no interference, greatly improving the anti-interference ability of the system. At the same time, the maximum ratio combining method of the MIMO antenna array further enhances the resistance to interference and improves the quality and reliability of the signal. Secondly, the use of long short-term memory network, a deep learning technology, enables the detection node to automatically learn and predict the channel state, realizing intelligent signal processing and optimization. Compared with traditional manual intervention or simple rule judgment, it can more accurately adapt to the complex and variable DDGS feed yard environment, improving the adaptability and intelligence of the system.

[0091] In step 2 of the above embodiment, the blind source separation algorithm constructs a cost function (Formula 1) through the generalized information maximization method, the core principle of which is to maximize the statistical independence of signal components to solve the problem of multi-path signal aliasing. The demixing matrix The signal matrix after pre-whitening processing is optimized, and the pre-whitening processing aims to eliminate the second-order correlation between signals (diagonalization of the covariance matrix), thereby reducing the complexity of subsequent separation. The objective function Usually a nonlinear function (such as the hyperbolic tangent function) is used to approximate the probability density distribution of independent components by enhancing signal sparsity. The regularization coefficient is used to prevent numerical instability problems caused by singular values of the matrix during gradient descent. The introduction of the feature vector matrix allows the algorithm to adaptively extract principal components from the observed signals, thereby improving separation efficiency.

[0092] In step S3, the gradient descent method iteratively updates the demixing matrix minimizing the cost function (formula 1), and finally obtaining the independent component matrix The selection of the main path signal is based on signal energy or signal-to-noise ratio (SNR) threshold, and the direct path signal (main path) and non-direct path signal (multipath interference) are determined by analyzing the time domain characteristics (such as envelope fluctuation) or frequency domain characteristics (such as power spectral density) of each component. This process is achieved by matrix eigenvalue decomposition, and the main path signal corresponds to the maximum eigenvalue component.

[0093] In steps S4 to S6, the construction of the multi-path channel propagation model depends on the three-dimensional coordinates of the detector , and the coordinates of the reflection points , The actual propagation path of the signal in space is calculated by the path propagation distance formula (formula 2). The path correction factor \beta takes into account the signal diffraction effect (such as Fresnel zone diffraction), and the path length is corrected by the geometric diffraction theory (GTD). The phase shift calculation formula (formula 3) combines the signal wavelength and the nonlinear weight , which quantifies the phase difference caused by the difference in propagation distance of different path signals. The generalized optimal weight method (formula 4) further introduces the path loss weight and the correction factor , which dynamically adjusts the contribution weight of each path signal through the path loss index, and finally realizes the comprehensive optimization of phase compensation.

[0094] Step S7 modifies the received signal in real time by the phase compensation factor (formula 6), where is calculated by formula 3 and formula 4. The channel nonlinear correction parameter is used to compensate for the phase distortion caused by the Doppler effect or channel time-varying characteristics during signal transmission, ensuring that the compensated signal can accurately reflect the original signal characteristics.

[0095] In actual implementation of step 2, the hardware of the detection node also needs to have a storage module for storing the received wireless signals to form an observation signal matrix on the basis of step 1, and a special signal processing chip or module to run algorithms such as blind source separation algorithm, multipath channel propagation model calculation, phase offset calculation, comprehensive phase calculation and phase compensation operation. In addition, a high-precision three-dimensional coordinate measurement device (such as a GPS module combined with an inertial measurement unit IMU) is needed to obtain accurate three-dimensional coordinate information of the detector. Among them, the signal processing chip or module is connected with the wireless communication module to receive the wireless channel data optimized after step 1. The storage module is connected with the signal processing chip or module for storing the observation signal matrix and other intermediate calculation results. The three-dimensional coordinate measurement device is connected with the signal processing chip or module to provide real-time three-dimensional coordinate information of the detector. The signal processing chip or module finally transmits the signal compensated after phase compensation back to the wireless communication module for subsequent data transmission.

[0096] In implementation, the specific workflow is: the sensor network collects the temperature and humidity, pressure data in the feed bin, and transmits them to the detection node through ZigBee; then, the detection node receives wireless signals through the MIMO antenna and stores them as an observation signal matrix X; the FPGA calls the pre-whitening module to perform decorrelation processing on X, and the ARM processor executes the gradient descent algorithm to optimize the demixing matrix W to extract the main path signal; then, the UWB positioning module updates the detector coordinates in real time, calculates the phase offset combined with the path propagation model (formulas 2-5), and generates the compensation factor; the compensated signal is uploaded to the central control platform through LoRaWAN, and the platform dynamically adjusts the working mode of the detection node according to the signal quality.

[0097] In implementation, the detection node completes self-organization through UWB positioning and establishes a three-dimensional coordinate mapping relationship; wireless signals are collected every 5ms, blind source separation and phase compensation are performed, and compensation parameters are updated; when the path loss index exceeds the threshold, the path reselection algorithm is triggered, and the standby antenna channel is switched to.

[0098] To verify the effectiveness of the blind source separation and phase compensation algorithm used in step 2 of the present application, comparative experiments were conducted under the same experimental conditions. Experimental group A used the present application technology, i.e. using the generalized information maximization criterion to construct a blind source separation cost function, then optimizing the demixing matrix through the gradient descent method to extract the main path signal, combining the three-dimensional coordinates of the detector to construct a multipath model, and performing phase weighted compensation through maximum ratio synthesis; while experimental group B used statistical separation technology, only using mean filter to process multipath effect, without phase compensation. Both groups conducted five independent experiments in a fixed test environment, and the measured parameters included signal-to-noise ratio (SNR, dB), bit error rate (BER, %), phase deviation (°) and data acquisition accuracy (%). The experimental record is shown in Table 1:

[0099] Table 1 Blind source separation and phase compensation algorithm experiment record table

[0100] Experimental group Number of experiments SNR (dB) Bit error rate (%) Phase deviation (°) Data acquisition accuracy (%) Group A 1 28.5 0.15 2.1 96.2 Group A 2 29.0 0.14 2.0 96.8 Group A 3 28.8 0.16 2.2 96.5 Group A 4 29.2 0.13 2.0 97.0 Group A 5 29.0 0.15 2.1 96.9 Group B 1 25.0 0.25 4.5 91.0 Group B 2 24.8 0.26 4.7 90.5 Group B 3 25.2 0.24 4.6 91.2 Group B 4 25.0 0.27 4.8 90.8 Group B 5 24.9 0.25 4.7 91.0

[0101] During the experiment, the detection node first forms an observation matrix after pre-processing the multipath signal collected, realizes channel state prediction through the LSTM network, and decomposes the data by the blind source separation module. The blind source separation algorithm used in group A is based on the generalized information maximization criterion, and the optimal solution mixing matrix is obtained through iterative optimization, and then a multi-path propagation model is constructed combined with three-dimensional position data to realize accurate phase compensation. Group B only uses traditional mean filtering processing and does not effectively compensate the multipath phase. The data of each group is realized through high-speed DSP for real-time signal processing and weighted synthesis, and finally the compensated signal is output for measurement and statistics. During the experiment, each parameter is automatically recorded by the built-in sensor and signal acquisition module, and the data is uploaded to the central monitoring system for subsequent analysis.

[0102] The experimental results show that group A is better than group B in SNR, bit error rate, phase deviation and data acquisition accuracy, indicating that the blind source separation and phase compensation algorithm used in step 2 of the application can effectively extract the main path signal and eliminate multipath interference, so that the quality of the final output signal is significantly improved.

[0103] In step 3 of the above embodiment, the inertial navigation module is a system that calculates the position, velocity and attitude of the carrier based on Newtonian mechanics by measuring its acceleration and angular velocity. It mainly consists of an accelerometer and a gyroscope. The accelerometer is used to measure the acceleration of the carrier in three orthogonal axes, and through integration with respect to time, the velocity can be obtained, and then the position information can be obtained by integrating the velocity. The gyroscope is used to measure the angular velocity of the carrier around the three orthogonal axes, and through the integration of the angular velocity, the attitude angle of the carrier can be obtained. In the DDGS feed detection, the inertial navigation module is installed on the detection node to measure the motion state of the detection node in real time and provide basic data for determining the node position. However, due to the problem of error accumulation with time in the inertial navigation system, when used alone, the position accuracy will decrease after a long time.

[0104] UWB is a kind of carrierless communication technology, which transmits data by using nanosecond to microsecond non-sinusoidal narrow pulse. In terms of ranging, UWB calculates the distance by measuring the time of flight (ToF) of the signal between two nodes. Specifically, the sending node sends a UWB signal, and the receiving node records the time of arrival of the signal. According to the signal propagation speed (the speed of light), the distance between the two nodes can be calculated. UWB ranging has the advantages of high precision, strong anti-multipath ability, low power consumption, etc. In the DDGS feed detection, the adjacent detection nodes perform ranging through UWB, providing accurate information for determining the relative position of the nodes.

[0105] Federal filtering is a kind of distributed filtering algorithm, which fuses the results of multiple sub-filters to obtain more accurate and reliable estimation value. In this step, the federal filtering algorithm fuses the inertial navigation module and the UWB ranging data. The inertial navigation module provides autonomous navigation information of the detection node, while the UWB ranging data provides relative position information between nodes. Through the fusion of these two kinds of data, the federal filtering algorithm makes full use of their advantages, suppresses the error accumulation of the inertial navigation system, and improves the accuracy of position estimation. Specifically, the federal filtering algorithm divides the entire filtering process into multiple sub-filters, each sub-filter processes a part of data (such as inertial navigation data or UWB ranging data), and then fuses the results of each sub-filter through information distribution principle to obtain the final node absolute position estimation.

[0106] After obtaining the absolute position of the detection node, combined with the temperature, humidity and PH3 concentration data collected by the detection node, the geographic information system (GIS) technology or related data visualization algorithm is used to map these data to the spatial position, generating the temperature, humidity and PH3 concentration distribution heat map. The heat map represents the temperature, humidity and PH3 concentration at different positions by the depth of color, which intuitively shows the distribution of environmental parameters in the DDGS feed yard.

[0107] In actual implementation, the hardware of the detection node includes an inertial navigation module (accelerometer and gyroscope), a UWB ranging module, a data processing unit (such as a microprocessor or an embedded system), a temperature and humidity sensor, a PH3 concentration sensor, and a communication module (such as a wireless transceiver). In addition, a display device (such as a touch screen or a remote monitoring terminal) is also needed to display the generated temperature and humidity and PH3 concentration distribution heat map. 2. Hardware connection: the inertial navigation module, the UWB ranging module, the temperature and humidity sensor, and the PH3 concentration sensor are respectively connected to the data processing unit to transmit the collected data to the data processing unit for processing. The communication module is also connected to the data processing unit for receiving and sending data, realizing communication between the detection nodes and communication with the remote monitoring terminal. The display device is connected to the data processing unit through wired or wireless means to receive and display the generated heat map.

[0108] In implementation, the specific working method is as follows: the detection node first monitors the GPS signal strength in real time. When the GPS signal strength is lower than the preset threshold, the inertial navigation module starts to work, measures the acceleration and angular velocity of the detection node in real time, and transmits the data to the data processing unit. At the same time, the adjacent detection nodes measure the distance through the UWB ranging module, and transmit the ranging data to the data processing unit. The data processing unit runs the federated filtering algorithm to fuse the inertial navigation data and the UWB ranging data, and calculates the absolute position of the detection node. At the same time of calculating the position, the temperature and humidity sensor and the PH3 concentration sensor collect the temperature and humidity and PH3 concentration data in the environment in real time, and transmit the data to the data processing unit. The data processing unit integrates the position data and the environmental parameter data, generates a temperature and humidity and PH3 concentration distribution heat map using a corresponding algorithm, and sends the heat map to the display device through the communication module for display.

[0109] Based on the above hardware, the same hardware platform is selected to experiment on the algorithm. Experiment group A adopts the federated filtering algorithm based on IMU and UWB data in step 3 of the present application, and experiment group B adopts a traditional Kalman filter for single node positioning. The two groups conduct five experiments under the same environmental conditions, and the comparison parameters include: positioning error (m), positioning delay (ms), data consistency (%), channel stability (dB), and robustness index. During the experiment, the detection nodes collect their own data, calculate the absolute position of the node through their own algorithm, and automatically transmit the data to the cloud for statistics. The data records of each group are shown in Table 2:

[0110] Table 2 Application experiment record table of step 3

[0111] Experimental group Number of experiments Positioning error (m) Positioning delay (ms) Data consistency (%) Channel stability (dB) Group A 1 0.85 120 97.5 28.0 Group A 2 0.80 115 98.0 28.5 Group A 3 0.83 118 97.8 28.2 Group A 4 0.82 116 98.1 28.4 Group A 5 0.84 119 97.9 28.3 Group B 1 1.20 150 92.0 25.0 Group B 2 1.25 155 91.5 24.8 Group B 3 1.22 152 91.8 25.1 Group B 4 1.28 157 91.2 24.9 Group B 5 1.26 154 91.6 25.0

[0112] The experimental results show that the A group adopting the federal filter algorithm is significantly better than the traditional Kalman filter method (B group) in positioning accuracy, response delay, data consistency, channel stability and robustness, verifying the positive and beneficial effects of the step 3 in multi-sensor data fusion and high-precision positioning.

[0113] Compared with the prior art, in the DDGS feed detection method based on wireless transmission, the prior art may not accurately determine the position of the detection node in the case of poor GPS signal, resulting in mismatch between the environmental parameter data and the actual position. However, the step fully utilizes the autonomy of the inertial navigation and the high precision of the UWB ranging, effectively suppresses the error accumulation of the inertial navigation system, and improves the positioning accuracy, so that the temperature, humidity and PH3 concentration data can accurately correspond to the corresponding position. Secondly, the method can rely on the inertial navigation module and the UWB ranging technology for positioning when the GPS signal is insufficient, and is not limited by the GPS signal, thereby enhancing the adaptability and reliability of the system in complex environments. In addition, the generated temperature, humidity and PH3 concentration distribution heat map can intuitively show the environmental parameter distribution in the DDGS feed yard, which is easier for operators to understand and analyze than the traditional table or text form data. Through the heat map, operators can quickly find the temperature abnormal area and the area with too high PH3 concentration, and take timely measures to ensure the quality and safety of the DDGS feed.

[0114] In step 4 of the above embodiment, the BP neural network is a multi-layer feedforward network trained by the error backpropagation algorithm, which consists of an input layer, a hidden layer and an output layer. In this step, the BP neural network is used to build an error correction model. The core principle is to pass the input data to the output layer through forward propagation, calculate the error between the output result and the expected target, and then pass the error signal back to the input layer layer by layer through backpropagation, adjust the connection weights and biases between layers to minimize the error. During the training process, the network continuously learns the mapping relationship between the input data (such as PH3 sensor raw data, dust concentration and humidity data) and the output data (accurate PH3 concentration data), so as to accurately predict and correct new input data.

[0115] Digital filtering is used to process the PH3 sensor raw data, remove noise and interference signals, and improve the quality and reliability of the data. Common digital filtering methods include mean filtering, median filtering, Gaussian filtering, etc. The normalization algorithm is to convert data of different ranges and magnitudes to a unified scale for subsequent processing and analysis. In this step, the PH3 sensor raw data, dust concentration and humidity data are converted into a unified feature vector through the normalization algorithm, so that the data has comparability and consistency, which is conducive to the learning and training of the BP neural network.

[0116] Batch Normalization is a technique for normalizing the input data of each layer during the training process of a neural network. By normalizing the data of each mini-batch, it makes the mean of the data 0 and the variance 1, thereby accelerating the convergence speed of the network training, reducing the problems of gradient vanishing and gradient explosion. In the feature normalization layer, the batch normalization mechanism adjusts the distribution of each dimension of data, making it easier for the network to learn the characteristics of the data and improving the generalization ability of the model.

[0117] ReLU is a commonly used nonlinear activation function. In the nonlinear activation layer, the ReLU function performs nonlinear mapping on the linearly transformed hidden layer output data, setting the negative part to 0 and keeping the positive part unchanged. This can introduce nonlinearity, allowing the neural network to learn more complex function relationships and capture the nonlinear interference of dust and humidity on PH3 data, improving the model's expressive power.

[0118] Multi-layer Perceptron is a feedforward neural network composed of multiple neuron layers. In the fully connected fusion layer, the weighted superposition mechanism of the multi-layer perceptron is used to fuse the hidden features. Each neuron receives input from the previous layer and calculates the output through weighted summation, with the weights determining the importance of each input feature. Through the layer-by-layer processing of the multi-layer perceptron, different levels of hidden features can be fused to output comprehensive correction parameters, quantitatively reflecting the global mapping of environmental factors on PH3 measurement error.

[0119] In the actual implementation of step 4, the application hardware mainly includes a detection node (built-in PH3 sensor, temperature and humidity sensor, dust concentration sensor, microprocessor, wireless communication module), edge gateway (processor with strong computing power, storage device, wireless communication module). The detection node is used to collect PH3 concentration, temperature and humidity and dust concentration data, and transmit the data to the edge gateway. The edge gateway is responsible for running the error correction model constructed by the BP neural network, processing and correcting the data. Among them, the PH3 sensor, temperature and humidity sensor and dust concentration sensor in the detection node are connected with the microprocessor respectively, and the collected data is transmitted to the microprocessor for preliminary processing. The microprocessor is connected with the wireless communication module of the edge gateway through the wireless communication module, and the processed data is sent to the edge gateway. The processor of the edge gateway is connected with the storage device, and the storage device is used to store the BP neural network model, training data and other related information.

[0120] In the actual implementation of step 4, the detection node samples data through the PH3 sensor, temperature and humidity sensor and dust concentration sensor according to the preset sampling period. The microprocessor preliminarily processes the collected raw data, including digital filtering and other operations, and then sends the data to the edge gateway through the wireless communication module. After receiving the data, the edge gateway first standardizes the data through the digital filtering and normalization algorithm of the input layer, generates a unified feature vector, and outputs it to the feature normalization layer. The feature normalization layer adjusts the data distribution using the batch normalization mechanism, and then the data enters the linear transformation hidden layer for linear extraction. The non-linear activation layer uses the ReLU activation function to capture non-linear interference, and the fully connected fusion layer fuses each hidden feature through the weighted superposition mechanism of the multi-layer perception, and outputs the comprehensive correction parameter. The output correction layer implements real-time dynamic compensation on the original PH3 data based on the comprehensive correction parameter, and obtains the calibrated PH3 concentration data. If there is deviation between the compensated data and the expected target, the dynamic update layer uses the back propagation method and the Adam optimization strategy to continuously iterate and adjust the weight and bias, and optimizes the error correction model. Finally, the edge gateway transmits the calibrated PH3 data to the subsequent processing device or monitoring terminal through the wireless communication module.

[0121] In order to verify the positive effect of the BP neural network error correction model on the real-time compensation of PH3 sensor data, experiments are divided into two groups: Group A uses the BP neural network error correction model of the application for data compensation; Group B uses the traditional linear calibration method, which only uses fixed coefficients to correct PH3 data and cannot dynamically adapt to environmental changes. The two groups of experiments are carried out in the same DDGS yard environment, and the same hardware platform is used for continuous sampling experiments for 5 times. The main comparison of the two groups is the positioning error, correction deviation, SNR improvement, response delay and data consistency. During the collection process, the probe node acquires the original concentration data through the PH3 sensor according to the preset sampling period, and the environmental module collects dust and humidity data; these data are preprocessed by the edge gateway and input into the respective correction model. During the experiment, various indicators are automatically recorded by the built-in sensor, and after wireless transmission, they are summarized and counted by the cloud server to form an experimental record table, as shown in Table 3:

[0122] Table 3 Application experiment record table of BP neural network error correction model

[0123] Experimental group Number of experiments Measurement error (ppm) Correction deviation (ppm) SNR improvement (dB) Response delay (ms) Data consistency (%) Group A 1 1.2 0.8 6.5 110 98.0 Group A 2 1.1 0.7 6.8 108 98.3 Group A 3 1.3 0.9 6.4 112 97.8 Group A 4 1.2 0.8 6.7 109 98.1 Group A 5 1.1 0.7 6.6 107 98.2 Group B 1 2.0 1.8 4.2 140 93.5 Group B 2 2.1 1.9 4.0 142 93.0 Group B 3 2.2 2.0 4.1 141 92.8 Group B 4 2.0 1.8 4.3 139 93.3 Group B 5 2.1 1.9 4.2 140 93.0

[0124] The experimental results show that Group A using the BP neural network error correction model is significantly better than Group B using the traditional linear calibration in terms of measurement error, correction deviation, SNR improvement, response delay and data consistency. The average measurement error of Group A is reduced by about 45%, the correction deviation is reduced by about 55%, the SNR improvement is increased by about 2.4dB, the response delay is reduced by about 30ms, and the data consistency is more than 98%, while the data consistency of Group B is only about 93%. These data fully prove that the BP neural network dynamic compensation model can capture the nonlinear interference of environmental factors on PH3 sensor data in real time and compensate it accurately, thereby significantly improving the accuracy and real-time performance of PH3 data in the wireless transmission DDGS feed detection system.

[0125] In step 5 of the above embodiment, the edge gateway constructs a concentration distribution function based on the calibrated PH3 data, which essentially establishes a mathematical mapping relationship between spatial position and corresponding PH3 concentration. Through this function representation, the spatial distribution of PH3 concentration in the DDGS feed yard can be described at the mathematical level. In practical applications, the spatial position vector can be a three-dimensional coordinate, and by using the PH3 concentration data collected at different positions by the sensor, the function can be fitted, thereby providing a basis for subsequent analysis of concentration changes. The gradient direction obtained by taking the derivative of the concentration distribution function represents the direction in which the function value changes most rapidly. By iterating in the opposite direction of the gradient, the local minimum of the function can be found (in tracking the concentration diffusion path, this is the reverse process of finding the concentration peak). In each iteration, the environmental gradient is determined according to the concentration data change rate to calculate the next search direction, which is a dynamic adjustment based on concentration change. The concentration change rate reflects the trend of concentration change in space, and by analyzing this change rate, the search direction towards the concentration source can be more accurately determined. The setting of the convergence criterion is to determine whether the local concentration peak position has been found, i.e., whether the target state of the search has been reached. The local abnormal area coordinates and the position data output in step 3 may differ due to measurement errors, different measurement baselines, etc. The spatial alignment mechanism adjusts the local abnormal area coordinates to the same spatial reference system as the position data through certain transformations (such as translation, rotation, scaling, etc.), so that the two can be effectively fused.

[0126] The position data correction algorithm is mainly used to process errors in the position data. In actual measurement, the position data obtained through inertial navigation module, UWB ranging or other means may have certain errors. The position data correction algorithm analyzes and processes multiple node position data, estimates and corrects these errors using statistical methods or other mathematical models, and improves the accuracy of the position data.

[0127] In the actual implementation of step 5, the scheme relies on the vector edge gateway and the node detection overall monitoring network distributed at key positions in the DDGS yard. The node detection hardware includes PH3 sensors, temperature and humidity sensors, and wireless communication modules for data synchronization; at the same time, each node is equipped with a GPS module, an inertial navigation module, and a UWB ranging module for providing positioning data. Low-power wireless protocols such as LoRa or ZigBee are interconnected, and data is transmitted through high-speed SPI, I 2Ctransmission to embedded processor. Edge gateway is composed of multi-core processor, FPGA and DSP, pre-installed with software system of the present embodiment, wherein the data structure, centralized distribution function includes construction of gradient iteration, iterative algorithm and spatial data fusion module. In the specific implementation process, the detection node starts the PH3 center edge gateway according to the cycle. First, the data fusion module is used to construct the spatial concentration distribution function, then the local gradient is calculated by numerical derivation, and the reverse tracking algorithm based on gradient descent is started. The algorithm starts at the initial position, adjusts the search direction using the gradient information calculated in each iteration, and adjusts the concentration change to meet the convergence criterion to obtain the local abnormal area coordinates. The high-efficiency edge gateway adopts a built-in spatial layout module to package the coordinates and global positioning data from GPS, IMU and UWB for weighted aggregation to correct the position of the final abnormal area. All hardware modules are interconnected through standard interfaces (SPI, I 2 C, UART, LVDS), and real-time operating system (RTOS) is used for task scheduling between modules to ensure good operation of data-intensive processing, algorithm iteration and communication transmission.

[0128] Compared with the existing method which only relies on fixed threshold or single positioning algorithm, the present embodiment realizes accurate modeling and dynamic area positioning of PH3 concentration distribution through multi-sensor data fusion and reverse tracking algorithm based on gradient descent. Traditional methods often cannot accurately capture local concentration peaks, resulting in existence or misjudgment of abnormality. The present scheme uses reconstructed data collected in real time to construct partition distribution function, and obtains accurate concentration gradient information by solving derivation to realize fast convergence positioning of partition hotspots. In addition, through spatial partition and fast convergence positioning of partition, the algorithm accurately matches the local abnormal coordinates with the global data, greatly improving the accuracy of abnormal area marking. The hardware platform uses sensor experiment positioning low-delay communication interface to ensure high real-time and stability of data acquisition and transmission, further improving the overall speed response of the system.

[0129] In step 6 of the above embodiment, each probe node preprocesses the collected local data through the embedded encryption module, and calculates the local loss function gradient in the encrypted state of the data using homomorphic encryption or differential privacy mechanism. This process uses a privacy protection algorithm to inject noise and perturb data into the original PH3, temperature and humidity, dust, and other multi-dimensional data, ensuring that the calculated gradient accurately reflects the sensor error without revealing sensitive data. By using digital signatures and TLS protocols, the integrity and confidentiality of the data uploaded by each node are ensured, making secure multi-party computation (SMC) possible, thereby providing a reliable data input for the federated learning framework. At the local node, after calculating the local gradient based on the preset loss function (such as mean square error MSE), an improved local gradient descent algorithm is used to iteratively update the neural network weights. This algorithm adjusts the learning rate adaptively (e.g., using the Adam optimizer) and adjusts the momentum factor, ensuring that the model can quickly converge when facing nonlinear disturbances. The local model update process fully utilizes distributed computing resources to achieve low-latency updates and uses built-in verification mechanisms to verify the accuracy of each update, ensuring that the weight update data generated by each probe node has high precision and robustness. After receiving the locally protected gradient from each probe node, the cloud server uses a distributed weighted averaging strategy to fuse the local update data. This process combines Bayesian inference to model the prior noise and perform posterior filtering on each local gradient data, effectively reducing random noise interference during the fusion process. At the same time, by introducing a differential privacy mechanism to reprocess the aggregated results, it ensures that key features are preserved while resisting data leakage risks during global model updates. This weighted fusion process relies on the uncertainty measurement of each node's update data and uses a weight adjustment strategy to ensure that the fusion result accurately reflects the statistical characteristics of the multi-pile yard data. After cloud aggregation is complete, the stochastic gradient descent (SGD) or its variants (such as Adam, RMSProp) are used to iteratively optimize the global anomaly diagnosis model parameters. This update process uses dynamic learning rate and regularization strategies to gradually approach the optimal solution using the weighted fused gradient data. The update algorithm periodically evaluates the performance of the global model and automatically adjusts the hyperparameters using a callback function, allowing the global model to adapt to the distribution characteristics of abnormal data in different pile yard environments. The updated global model parameters are sent to the edge gateway through an encrypted feedback channel, providing the latest anomaly detection strategy to the probe nodes. After the updated global model parameters are fed back to each probe node through a secure transmission protocol, the node uses a local model fusion mechanism to match the global parameters with the pre-deployed local model. In this process, the local node uses a parameter fusion algorithm (such as weighted averaging or adaptive fusion algorithm) to fine-tune the global model based on the credibility and error indicators of the local data, thereby implementing individualized anomaly detection strategies.The fusion mechanism uses cross-validation and online calibration techniques to ensure the adaptability of model parameters in local environments, thereby improving the sensitivity and accuracy of anomaly detection. The entire federated learning process relies on secure multi-party computation (SMC) mechanisms. Data transmission between each detection node and the cloud is achieved through TLS, VPN, and secure communication protocols based on MQTT / CoAP, ensuring that no one party leaks original sensitive data during data aggregation, model updating, and parameter feedback. The security mechanism ensures data integrity and confidentiality during transmission and processing through dynamic key negotiation, digital signatures, and access control policies. This technology achieves privacy protection for data in multi-source data collaborative training while ensuring the stable operation of the federated learning framework, providing a solid security support for the system.

[0130] In the actual implementation of step 6, each node first calculates the gradient data under privacy protection locally and uploads the data after protection using encryption algorithms such as homomorphic encryption or differential privacy mechanisms. After receiving the data, the cloud uses distributed system architecture and efficient parallel computing to perform weighted averaging, Bayesian filtering, and noise reduction on the local update data, then iteratively optimizes the global model parameters using SGD, and feeds back the updated model parameters to the edge gateway through a secure interface. The edge gateway then fuses and adjusts the global model with the local model to achieve real-time optimization of the anomaly diagnosis strategy. Each module is interconnected through standard communication protocols such as MQTT, CoAP, etc., and low-latency networks to ensure overall system collaboration.

[0131] Compared with existing centralized anomaly detection methods, the embodiment fully utilizes the federated learning framework to achieve distributed collaborative training of multi-pile yard data, significantly reducing the risk of single-point data leakage. By using privacy-protected gradient calculation and secure multi-party computation at the local node, the security of data during transmission is ensured, and the robustness of data aggregation is improved using Bayesian inference and differential privacy technology. Global model updating through SGD and dynamic learning rate adjustment enables the model to quickly adapt to environmental changes and improve anomaly detection accuracy. Each node dynamically adjusts the detection strategy through local model fusion, ensuring the adaptability of the model in its own environment. Experimental data show that this scheme outperforms traditional methods in terms of positioning accuracy, response delay, and anomaly detection accuracy, effectively improving the real-time and stability of the DDGS pile yard monitoring system and enhancing the level of safety warning and risk management.

[0132] In step 7 of the above embodiment, game theory is a theory that studies decision-making when decision-makers directly interact and the equilibrium of such decisions. In this step, each probe node is considered as a participant in the game. Building the transmission quality and energy consumption utility functions is the core of this model. The transmission quality function mainly considers factors such as signal strength and bit error rate, as these factors directly affect the accuracy and reliability of data transmission; the energy consumption utility function focuses on battery power consumption to evaluate the energy efficiency of nodes at different power transmission. Through these two functions, the "benefit" of each node at different transmission power is comprehensively measured. Distributed iterative solution of Nash equilibrium means that without centralized control, each node finds a stable state, Nash equilibrium, by constantly interacting with other nodes and adjusting strategies. In the Nash equilibrium state, the strategy of each node is the optimal response to the strategies of other nodes, and at this time, any unilateral change in the transmission power strategy of a node will not increase its "benefit". Through this way, the optimal transmission power allocation scheme is finally obtained, which makes the entire system ensure transmission quality while minimizing energy consumption as much as possible.

[0133] LoRa (Long Range) is a kind of ultra-long distance wireless transmission technology based on spread spectrum technology. Its working principle is to use spread spectrum modulation technology to expand the bandwidth of the original signal to a wider frequency band for transmission. This technology has the characteristics of low power consumption and long distance transmission. In terms of low power consumption, LoRa reduces the energy consumption of devices by reducing data transmission rate and other ways, which is suitable for battery-powered devices. In terms of long distance transmission, LoRa can achieve long communication distance at relatively low power, which makes it have advantages in some scenarios that require communication distance and do not require high data transmission rate. When the probe node's power is less than 20% and it is in a non-exceptional area, switching to LoRa communication mode is to take advantage of its low power consumption and long distance transmission characteristics to prolong the working time of the node, while ensuring that data can be effectively transmitted.

[0134] In the actual implementation of step 7, the application hardware mainly includes a detection node (built-in battery power detection module, communication load detection module, power adjustment module, LoRa communication module, conventional wireless communication module, microprocessor) and an edge gateway (a device with data processing and communication functions). The battery power detection module is used to monitor the battery power in real time, the communication load detection module is used to detect the current communication load, the power adjustment module is used to adjust the transmission power according to the power allocation result, the LoRa communication module and the conventional wireless communication module are used for data transmission, and the microprocessor is responsible for controlling the operation of each module and data processing. Among them, the battery power detection module and the communication load detection module are connected with the microprocessor, and the detected data is transmitted to the microprocessor. The microprocessor is connected with the power adjustment module, and controls the power adjustment module to adjust the transmission power according to the power allocation result. The microprocessor is connected with the LoRa communication module and the conventional wireless communication module at the same time, and selects the appropriate communication module for data transmission according to different situations. The detection node communicates with the edge gateway through the wireless communication module, and sends the collected data to the edge gateway, and receives the instructions of the edge gateway at the same time.

[0135] In the implementation, the battery power detection module of the detection node monitors the battery power in real time, and the communication load detection module detects the current communication load data in real time, and transmits these data to the microprocessor. The microprocessor sends the battery power and the communication load data to the edge gateway. After receiving the data, the edge gateway uses the game theory power allocation model, takes each detection node as a game participant, constructs the transmission quality and energy consumption utility function, solves the Nash equilibrium through distributed iteration, calculates the optimal transmission power of each node, and sends the power allocation result to the microprocessor of the detection node. The microprocessor controls the power adjustment module to adjust the transmission power according to the received power allocation result. In the running process, the microprocessor constantly monitors the battery power. If the node power is lower than 20% and the node is in a non-exceptional area (the state information of the area can be obtained by interacting with the edge gateway), the microprocessor controls the switch to the LoRa communication mode, and uses the LoRa communication module for data transmission; if the above conditions are not met, the conventional wireless communication module is used for data transmission.

[0136] Compared with the traditional fixed power allocation method, the scheme can respond to the node battery state and communication load changes in real time, construct the transmission quality and energy consumption utility function, use distributed iterative solution Nash equilibrium to obtain the optimal energy consumption condition of each node transmission power configuration, effectively reduce the node energy consumption, and prolong the device endurance time. Secondly, by automatically switching the low power node to the LoRa communication mode, the scheme ensures that the low power communication link can be maintained under the condition of insufficient battery power, avoiding the monitoring interruption caused by insufficient power. Furthermore, the method uses multi-node collaborative calculation to realize the optimal power allocation in the whole network by using the distributed game model, significantly reduces the system internal interference and bit error rate, and improves the overall transmission quality and data acquisition accuracy. The experimental results show that the scheme can reduce the node energy consumption by more than 30% in the actual DDGS yard application, improve the communication link stability by about 2-3dB, and significantly reduce the overall system response delay, ensuring the real-time and accuracy of environmental monitoring data, and providing reliable and intelligent technical support for the safety management and abnormal warning of DDGS yard.

[0137] In step 8 of the above embodiment, periodic detection is a time-driven strategy that comprehensively evaluates the health status of the detection node by setting a fixed time interval. For battery power, it is essentially to monitor the energy reserve of the node, because the battery power is directly related to whether the node can continue to work normally. Low battery power will cause the node to be unable to complete data collection and transmission tasks, just like a machine that has lost power. Sensor data drift is caused by factors such as long-term use of sensors and environmental changes, resulting in a deviation between the sensor output data and the true value. This drift may lead to misjudgment of the DDGS feed status, affecting the accuracy of the entire monitoring system. The detection of wireless communication state mainly focuses on signal strength, bit error rate and other indicators. Wireless communication is the bridge for nodes and other devices to interact with data, and poor communication state will cause data transmission delay and loss, affecting the real-time and reliability of the system. The preset abnormal threshold is a boundary value set according to the design requirements and actual experience of the system. When the detection data exceeds this threshold, it indicates that the node has an abnormality in some aspects, which may affect the normal operation of the system.

[0138] When the data of a node is detected to exceed the preset abnormal threshold, the node is marked as a faulty node, which is a mechanism for quickly identifying and locating problems. The marked faulty node is like a "problem point" with a label, facilitating subsequent processing. Adjacent node topology reconstruction is to ensure the connectivity of the entire monitoring network and the smoothness of data transmission. In a wireless sensor network, nodes cooperate with each other to collect and transmit data. When a node fails, its adjacent nodes need to adjust the connection relationship and form a new topology. This is like a traffic network where a road is out of service, and the surrounding roads need to be re-routed to ensure smooth traffic.

[0139] The early warning control host sends a state query instruction to adjacent nodes through a broadcast protocol. The broadcast protocol is an efficient information dissemination method that can send information to multiple nodes in the network simultaneously. In this scenario, by broadcasting the state query instruction, the latest location information, communication quality parameters, and network connection state of adjacent nodes can be quickly collected. These information are the basis for subsequent topology reconstruction and route update, just like the need to understand the surrounding road conditions before repairing the road.

[0140] The shortest path method is a classic graph algorithm used to find the shortest path between two nodes in a graph. In this method, the node connection weight is composed of node remaining energy, signal quality, and link delay, which comprehensively considers the energy consumption, communication reliability, and data transmission speed of the node. By calculating the node connection weight, each connection can be assigned a "cost". The shortest path method is to find the path with the minimum cost. When a faulty node appears in the network, the original route may no longer be optimal, and dynamic route update needs to be triggered to ensure that data can be transmitted through the best path. This is like a navigation system that will re-plan the route and choose the fastest road when there is traffic congestion.

[0141] In actual implementation of step 8, the hardware mainly includes detection nodes, early warning control host and communication network equipment. The detection node is composed of a battery, various sensors (such as temperature and humidity sensors, PH3 sensors, etc.), a wireless communication module and a microprocessor. The battery provides energy for the node, the sensors are responsible for collecting relevant data, the wireless communication module is used for communication with other nodes and the early warning control host, and the microprocessor is responsible for controlling the operations and data processing of the node. The early warning control host is usually a high-performance computer equipped with large-capacity storage devices and high-speed processors, which is used to receive, process and analyze the data of the detection nodes, and issue control instructions. The communication network equipment includes routers, switches and the like, which are used to build a wireless sensor network to ensure smooth communication between nodes and between nodes and the early warning control host. Among them, the battery in the detection node is connected with the microprocessor and the wireless communication module to provide power support for them. Various sensors are connected with the microprocessor to transmit the collected data to the microprocessor for processing. The microprocessor is connected with the communication network equipment through the wireless communication module to send the processed data to the early warning control host and receive the instructions of the early warning control host. The early warning control host establishes communication connection with each detection node through the communication network equipment to realize data interaction and control.

[0142] In implementation, the detection node detects its battery capacity, sensor data drift and wireless communication state according to the preset period, and sends the detection results to the early warning control host through the wireless communication module. After receiving the detection data, the early warning control host compares it with the preset abnormal threshold. If it is found that any data exceeds the preset abnormal threshold, the early warning control host marks the node as a fault node, and sends a state query instruction to the neighboring nodes of the fault node through a broadcast protocol. After receiving the instruction, the neighboring nodes send their latest location information, communication quality parameters and network connection state to the early warning control host. The early warning control host calculates the node connection weight according to the received information, and calculates the best route using the shortest path method. Finally, the early warning control host triggers dynamic route update in the network, and synchronously reports maintenance instructions to notify relevant personnel to repair or replace the fault node.

[0143] Compared with the traditional fixed routing and centralized monitoring method, the scheme of the embodiment has the following beneficial effects: first, through multi-parameter real-time monitoring, the overall detection of the health status of the detection node is realized, and problems such as insufficient battery power, sensor drift or communication anomaly can be found in time, thereby reducing the risk of system interruption caused by single node failure; second, the fault determination mechanism based on decision tree and multi-dimensional statistics is adopted, the accurate classification and labeling of abnormal state are realized, and reliable basis is provided for subsequent dynamic topology reconstruction; third, the broadcast query and distributed data fusion algorithm are used, the neighbor node information can be quickly collected by the early warning control host, the best route is determined through the shortest path method and the minimum spanning tree algorithm, the network self-healing is realized, and the robustness and communication continuity of the whole system are significantly improved; finally, the scheme fully utilizes low-power and high-performance sensor and processor modules on the hardware, realizes efficient cooperation between modules through standard interface and secure communication protocol, and ensures the real-time performance of data transmission and processing. The experimental results show that the scheme can improve the node fault detection accuracy by about 25%, shorten the dynamic routing update response time by 30%, and significantly improve the overall stability of the network, thereby providing strong technical support for intelligent monitoring and risk early warning of the DDGS yard.

[0144] Although the specific embodiments of the present application are described above, those skilled in the art should understand that these specific embodiments are only illustrative, and those skilled in the art can make various omissions, replacements and changes to the details of the above-mentioned method and system without departing from the principles and essence of the present application. For example, the above-mentioned method steps are combined, and the substantially same function is performed in a substantially same method to achieve a substantially same result, which belongs to the scope of the present application. Therefore, the scope of the present application is only limited by the appended claims.

Claims

1. A DDGS feed detection method based on wireless transmission, characterized in that: Includes the following steps: Step 1: Based on the real-time collected RSSI and bit error rate data, the probe node predicts the channel attenuation trend through a long short-term memory network. If the predicted signal quality is lower than a preset threshold, the wireless operating frequency band is switched through an adaptive frequency hopping mechanism, and the real-time signal is weighted and synthesized through a MIMO antenna array. Step 2: Based on the optimized wireless channel data from Step 1, the detection node extracts the main path signal using the blind source separation algorithm and calculates the phase offset of the signals on different paths by combining the detector location topology data, and performs phase compensation on the received signal. Step 3: When the GPS signal strength is detected to be lower than the preset threshold, the detection node uses the inertial navigation module and the UWB ranging data of the adjacent nodes to generate the absolute position of the node by using the federated filtering algorithm, and outputs a heat map of temperature, humidity and PH3 concentration distribution. Step 4: The detection node samples data through the PH3 sensor according to the preset sampling period, and transmits the collected data and the location information output in Step 3 to the edge gateway through the wireless channel. The edge gateway uses a BP neural network to build an error correction model, performs real-time compensation on the PH3 sensor data based on the dust concentration and humidity data in the current environment, and outputs the calibrated PH3 data. Step 5: If the PH3 concentration at the same detection node exceeds the preset threshold for three consecutive samplings and the temperature and humidity meet the characteristics of fermentation anomalies, the concentration diffusion path is traced in reverse using the gradient descent algorithm. The abnormal core area is marked by combining the location data output in Step 3, and the abnormal data is output to the early warning control host. Step 6: The early warning control host uploads the locally collected data to the cloud. The cloud aggregates data from multiple storage yards through a federated learning framework to train an anomaly diagnosis model and feeds back the updated model to the detection node. Step 7: Collect battery power and current communication load data for each detection node, allocate transmission power through a game theory power allocation model, and switch to LoRa communication mode if the node's battery power is below 20% and it is not in an abnormal area. Step 8: Periodically check the health status of the probe nodes, including battery level, sensor data drift, and wireless communication status; if any data exceeds the preset abnormal threshold, mark it as a faulty node and initiate topology reconstruction of adjacent nodes, and simultaneously report maintenance instructions.

2. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: The working method of step 1 is as follows: Based on the real-time collected RSSI and bit error rate data, the probe node first inputs the collected channel state data as a feature vector into the long short-term memory network, and constructs a channel attenuation model in the time series prediction process to predict the future signal quality; During model training and prediction, weights are optimized using historical data to obtain the predicted signal attenuation trend curve. If the prediction results indicate that the signal quality is lower than a preset threshold, an adaptive frequency hopping mechanism is adopted. The working frequency band is analyzed and selected in real time using energy detection and interference identification algorithms, and spectrum switching is performed using a software-defined radio interface. Based on the multi-channel data received after switching, the maximum ratio synthesis method is used to weight and spatially filter the signals of each channel through a MIMO antenna array, and the synthesized signal is output as the optimal channel data.

3. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: The steps in step 2 include: Step S1: Based on the optimized wireless channel data from Step 1, the probe node receives and stores wireless signals containing multipath effects to form an observation signal matrix. , ;in Indicates the number of receiving antennas; Indicates the number of sampling points; Step S2: Analyze the observed signal matrix using a blind source separation algorithm. Preprocessing is performed, and a cost function is constructed based on the generalized information maximization method. To achieve statistical independence of each independent component, the formula for the cost function is as follows: （1） In formula (1), For the unmixing matrix, This is the pre-whitened signal matrix, used to reduce signal correlation; The objective function is used to enhance signal sparsity control. This is the regularization coefficient, used to prevent numerical overflow; The received observation signal matrix eigenvector matrix; Step S3: Based on the cost function The unmixing matrix is ​​iteratively optimized using the gradient descent method. The optimal independent component matrix is ​​obtained. And select the main path signal from it. ; Step S4: Based on the detector's three-dimensional coordinates , Construct a multipath channel propagation model and define the signal propagation path. Propagation to scattering point , propagation distance for: （2） In formula (2), , For path Spatial coordinates of the reflection point; , The coordinates of the signal source; This is the path correction factor, used to account for signal diffraction effects; For path The angle of incidence; Step S5: Calculate the phase shift of the signal along each path using the propagation model. The calculation formula is: （3） In formula (3), Indicates the signal wavelength. For reference position coordinates; It is a nonlinear phase compensation weight used to correct the interference effects of signals from different paths; Step S6: Considering the attenuation characteristics of signals along each path, the generalized optimal weighting method is used to calculate the composite phase. The calculation formula is: （4） In formula (4), For path loss weights; This is the path loss exponential correction factor, where This is the loss adjustment coefficient; The number of received paths; where the path loss weights are... In the middle, path loss value The calculation formula is: （5） In formula (5), , The path decay exponent; As a correction factor; Step S7: During real-time processing, the detection node adjusts its position according to the phase shift. For the received signal Perform a compensation operation to obtain the compensated signal. The formula for the compensation operation is: （6） In formula (5), For phase compensation factor, Here, is the remote channel correction factor, where These are the channel nonlinearity correction parameters.

4. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: In step 3, the detection node first collects the acceleration data output by the inertial navigation module. Angular velocity data And the relative distance data obtained by adjacent nodes through the UWB ranging module State vectors are constructed using the federated filtering algorithm. ,in Indicates node position, Indicates speed, Indicates the number of sampling points; And set the state transition matrix Control input matrix Covariance of process noise The prior state estimate of each node is obtained using the prediction formula and the prediction error covariance, and prediction is performed for each local node; the prediction formula is: （7） The prediction error covariance is: （8） In formulas (7) and (8), This represents the acceleration input obtained from the IMU. This represents the posterior state estimate from the previous time step; Indicates process noise; for The error covariance matrix; then, in the local update phase, the observation vectors obtained by the inertial navigation module and the UWB ranging module are combined. Define the observation matrix Covariance of measurement noise Calculate the Kalman gain and update the local posterior state estimate and posterior error covariance matrix. The formula for calculating the Kalman gain is: （9） In formula (9), This is the observation matrix, used to map the state vector to the measurement space; For covariance; For measuring noise; For those participating in federated filtering For each node, a weighted optimal fusion method is used to obtain the global state estimate by combining the state estimates and corresponding covariances of each local node. The formula expression is: （10） In formula (10), For the first Local posterior state estimation of nodes; This is the corresponding error covariance; The total number of received paths is calculated. Finally, the first three dimensions of the global state estimate are extracted as the absolute positions of the nodes, and the temperature, humidity and PH3 concentration data corresponding to each node are correlated with the absolute positions of the nodes using the Kriging interpolation method to generate a heat map of the distribution of environmental parameters in the DDGS stockpile.

5. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: The error correction model includes an input layer, a feature normalization layer, a linear transformation hidden layer, a nonlinear activation layer, a fully connected fusion layer, an output correction layer, and a dynamic update layer. The input layer is used to standardize the raw data from the PH3 sensor and the dust concentration and humidity data collected by environmental detection through digital filtering and normalization algorithms, generate a unified feature vector, and output it to the feature normalization layer. The feature normalization layer is used to adjust the data distribution of each dimension using a batch normalization mechanism; the linear transformation hidden layer is used to linearly extract sensor drift data using a weight matrix and a bias vector; the nonlinear activation layer is used to perform nonlinear mapping using the ReLU activation function to capture the nonlinear interference of dust and humidity on PH3 data; the fully connected fusion layer is used to fuse the hidden features through the weighted superposition mechanism of a multilayer perceptron and output a comprehensive correction parameter, which is used to quantitatively reflect the global mapping of environmental factors on PH3 measurement error; the output correction layer is used to perform real-time dynamic compensation of the original PH3 data based on the comprehensive correction parameter using element-wise multiplication to obtain calibrated PH3 concentration data; the dynamic update layer is used to continuously iteratively adjust the weights and biases using a backpropagation method and an Adam optimization strategy when there is a deviation between the compensated data and the expected target.

6. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: In step 5, the edge gateway first constructs a concentration distribution function based on the calibrated PH3 data. ,in Represents a spatial location vector. The value represents the position. The pH3 concentration at the location; then, the edge gateway uses a gradient descent algorithm to analyze the concentration distribution function. The system performs differentiation to obtain the concentration change rate and determine the direction of the local concentration gradient. It then starts iterative updates from the initial sampling position. In each iteration, based on the concentration data change rate, the system calculates the environmental gradient to determine the next search direction. It then uses a convergence criterion to determine whether the local concentration peak has reached the convergence condition. If so, it obtains the coordinates of the local anomaly region obtained by reverse tracing. If not, it continues iterative updates. Next, the edge gateway uses a spatial alignment mechanism and a position data correction algorithm to fuse the coordinates of the local anomaly region with the position data output in step 3. During the data fusion process, a weighted average method is used to statistically process the position data of each node.

7. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: The working method of step 6 is as follows: After the early warning control host uploads the local data collected by each distributed detection node to the cloud server through a secure transmission protocol, the cloud starts the federated learning framework and asynchronously aggregates the local model updates from multiple DDGS storage farms through a secure multi-party computation mechanism. During the aggregation process, each detection node first calculates the local loss function gradient locally using a privacy protection mechanism, generates local model weight update data through a local gradient descent algorithm, and transmits it to the cloud. After receiving multiple local gradients, the cloud uses a weighted average strategy to fuse the local update data and uses Bayesian inference and differential privacy algorithms to filter and denoise the gradient data. Then, the cloud iteratively updates the global anomaly diagnosis model parameters through stochastic gradient descent and feeds the updated model parameters back to each detection node. Finally, the detection nodes match the updated global parameters with the local model through a local model fusion mechanism and adjust the local anomaly detection strategy.

8. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: The game theory power allocation model treats each node as a player in the game. By constructing transmission quality and energy consumption utility functions, it uses distributed iterative solution to solve the Nash equilibrium to obtain the optimal transmission power allocation.

9. The DDGS feed detection method based on wireless transmission according to claim 1, characterized in that: In step 8, when any data exceeds the preset abnormal threshold, the node status is updated to a faulty node. Based on the node tag, the early warning control host sends a status query command to neighboring nodes through a broadcast protocol to collect the latest location information, communication quality parameters and network connection status of each node. Then, the early warning control host calculates the best route according to the node connection weight using the shortest path method and triggers dynamic route updates in the network. The node connection weight is composed of the node's remaining energy, signal quality and link delay.

10. A DDGS feed detection system based on wireless transmission, characterized in that: The DDGS feed detection method based on wireless transmission, as described in any one of claims 1-9, includes: an early warning control host and a distributed wireless detection and rodent repellent integrated machine; The early warning control host includes a host computer system, a first wireless communication module, a channel evaluation and frequency hopping module, and a remote management and control interface; the distributed wireless detection and rodent repelling integrated machine includes a second wireless communication module, an inertial navigation module, a temperature and humidity detection module, a pH3 sensor module, a dust sensor module, an ultrasonic rodent repelling module, a MIMO antenna array, and a lithium battery module; the output terminals of the inertial navigation module, temperature and humidity detection module, pH3 sensor module, and dust sensor module are connected to the host computer system of the early warning control host through the second wireless communication module; the output terminal of the host computer system is connected to the inertial navigation module, temperature and humidity detection module, pH3 sensor module, dust sensor module, ultrasonic rodent repelling module, MIMO antenna array, and lithium battery module of the distributed wireless detection and rodent repelling integrated machine through the first wireless communication module.

Images