Distributing device for silo and control system

Abstract

The invention relates to the technical field of silo material storage and conveying, in particular to a silo material distribution device and a control system.The silo material distribution device comprises a material flowing monitoring module, an execution module, a control module and an abnormity alarm module; the control system comprises a data acquisition unit, a data processing unit, an analysis and prediction unit, a parameter optimization unit, an execution control unit, a storage unit and a remote interaction unit. During working, material flow states and characteristic data are collected through multiple types of sensors, after filtering and anomaly classification processing, the flow states are classified through a support vector machine model, the discharging trend is predicted through a BP neural network, vibration parameters (frequency, amplitude and time sequence) and the material distribution chute angle are optimized based on a linear regression model, and the material distribution chute angle is obtained. And closed-loop control of monitoring, processing, prediction, optimization, execution and feedback is formed. According to the invention, material characteristics and flow states can be accurately associated, cooperative achievement of blocking prevention, uniform material distribution and multi-material accurate proportioning is realized, complex working condition changes are adapted, control accuracy and production continuity are improved, remote centralized management and control are supported, and manual intervention cost is reduced.

Description

Technical Field

[0001] This application relates to the field of silo material storage and conveying technology, and in particular to a silo material distribution device and control system. Background Technology

[0002] In industrial production, silos are crucial equipment for storing and transporting materials, and the precision of their material distribution devices directly impacts the accuracy of multi-material mixing and the improvement of production efficiency. Especially in scenarios involving the mixing of multiple materials, ensuring that each material is uniformly distributed according to a predetermined ratio becomes a key aspect of guaranteeing product quality and process stability. Research in this area not only affects production costs but also plays a decisive role in the smoothness of downstream processing, making its importance self-evident.

[0003] However, current technologies in the field of silo feeding systems still have significant shortcomings. Many methods struggle to adapt to the differences between various materials with complex properties, leading to uneven distribution or blockages during the discharge process. Especially when handling powdery or granular materials, existing technologies lack the ability to dynamically adjust the material flow state, making it difficult to cope with unexpected situations caused by changes in material properties. This poses a significant challenge to achieving precise proportioning. Focusing further on the technical difficulties, the core issue for silo feeding devices in achieving intelligent proportioning of multiple materials lies in the effective control of material flowability.

[0004] Material flowability is affected by a variety of factors, among which setting vibration parameters is the primary challenge. The physical properties of different materials, such as particle size and moisture content, require flexible adjustment of vibration frequency and amplitude; otherwise, material may accumulate at the bottom of the silo or form voids, affecting the continuity of discharge. This problem is directly related to another crucial factor: the synergy between vibration and proportioning control. If the vibration device cannot match the discharge flow rate requirements in real time, it will be too vigorous at low flow rates, causing material dispersion, or insufficient at high flow rates, causing blockages. For example, in a chemical production scenario, the silo needs to process multiple particulate materials simultaneously. Improper vibration settings can lead to proportioning imbalances or even production interruptions.

[0005] Therefore, how to dynamically optimize vibration parameters in silo material distribution devices and ensure seamless linkage between vibration and proportioning control has become a key issue in achieving intelligent and precise material distribution for multiple materials. To address this, a material distribution device and control system for silos has been invented to solve the problems mentioned in the background technology. Summary of the Invention

[0006] In order to provide a silo fabric distribution device and control system.

[0007] This application provides a silo feeding device and control system, which adopts the following technical solution: including: The material flow monitoring module includes multiple sensors distributed in multiple locations within the silo, used to collect flow state data and material characteristic data of the material within the silo. The flow state data includes material flow velocity and distribution data, and the material characteristic data includes material particle size and moisture content. The execution module includes a fabric chute, a chute angle adjustment unit for driving the fabric chute, and a vibration mechanism acting on the silo. The vibration mechanism is equipped with independently adjustable vibration parameters, including vibration frequency, amplitude, and action sequence. The control module is communicatively connected to the material flow monitoring module, the chute angle adjustment unit, and the vibrating mechanism, respectively. It is used to receive and process the flow state data and material characteristic data, and generate control commands based on the processing results to synchronously adjust the tilt angle of the fabric chute and the vibration parameters of the vibrating mechanism.

[0008] Optionally, the sensors include a flow sensor, a density sensor, a level sensor, and a near-infrared spectral sensor; the flow sensor is arranged at the silo inlet, the inlet end and the outlet end of the fabric chute; the density sensor is symmetrically arranged on both sides of the middle of the fabric chute; the level sensor is distributed on the silo wall; and the near-infrared spectral sensor is arranged below the inlet.

[0009] Optionally, the vibration mechanism includes multiple vibration motors, which are circumferentially distributed to different areas of the lower conical section of the silo, and the vibration parameters of each vibration motor can be independently controlled.

[0010] Optionally, an abnormality alarm module is also included. The abnormality alarm module is communicatively connected to the control module and is used to trigger audible, visual, and / or remote alarm signals when the material flow monitoring module detects a flow abnormality or when the parameters in the control command exceed a preset safety range.

[0011] A material distribution control system for silos includes: a data acquisition unit, used to acquire flow status data and material characteristic data from the material flow monitoring module in real time; The data processing unit is used to filter and denoise the collected data, and to mark and classify abnormal data based on preset material characteristic thresholds, generating a subset of classified abnormal data. The analysis and prediction unit includes a pre-trained support vector machine model and a neural network model; the support vector machine model is used to classify and identify the material flow state based on the abnormal data subset; the neural network model is used to predict the material discharge continuity trend in future periods based on historical data and the current state. The parameter optimization unit is used to calculate the contribution of each element in the vibration parameters to the discharge stability through a regression analysis model based on the classification and identification results and the discharge continuity trend, and to generate an optimized combination of vibration parameters and an adjustment command for the angle of the material chute. The execution control unit is used to send the optimized vibration parameter combination and angle adjustment command to the execution module of the silo fabric device to drive its operation.

[0012] Optionally, the classification categories of the support vector machine model include at least normal state, blockage state, and uneven discharge state; the neural network model is a BP neural network model trained based on historical flow state data, vibration parameters, and actual discharge effect data.

[0013] Optionally, the parameter optimization unit is further configured to: calculate the real-time discharge fluctuation amplitude; when the fluctuation amplitude exceeds a set threshold, determine that the current state is an unstable state; analyze the contribution coefficients of vibration frequency, amplitude, and timing to the fluctuation under the unstable state based on a regression analysis model; generate multiple sets of candidate vibration parameter combinations based on the contribution coefficients; and select the combination that minimizes the predicted fluctuation amplitude as the optimized vibration parameter combination through simulation or historical data matching.

[0014] Optionally, a remote interaction unit is also included, which provides a visual monitoring interface to support remote status monitoring, parameter setting, centralized management and control of the fabric device, and receiving alarm information pushed by the abnormal alarm module.

[0015] Optionally, a remote interaction unit is also included, which is used to realize visual monitoring, remote parameter adjustment and abnormal alarm information push, and supports centralized management and control of multi-silo material distribution devices.

[0016] In summary, this application includes the following beneficial technical effects: 1. More accurate monitoring: By directly collecting material flow status and characteristic data through distributed multi-type sensors, flow difference indicators are generated. Compared with traditional indirect current monitoring, it can more accurately identify the root cause of abnormal material flow. 2. Smarter Control: It integrates support vector machine and neural network algorithms to achieve flow state classification and material discharge trend prediction. It combines linear regression model to optimize vibration parameter combination and control covers three dimensions: frequency, amplitude and time sequence, adapting to the dynamic changes of material characteristics. 3. More comprehensive functions: Simultaneously achieves anti-blocking, uniform material distribution and precise proportioning of multiple materials, solving the limitations of single-function control solutions and reducing the risk of production interruption and product quality fluctuations; 4. More convenient operation: Supports remote visual monitoring and centralized management, and is equipped with an anomaly alarm module to reduce manual intervention and improve operation and maintenance efficiency.

[0017] Instruction manual illustrations Figure 1 This is a logic architecture diagram of the control system for this device; Figure 2 This is a flowchart of the working process of the silo feeding device and control system of this unit; Figure 3 This is a flowchart illustrating the collaborative workflow of the artificial intelligence algorithms in this device. Figure 4 This is a schematic diagram of the structural connections of this device; Detailed Implementation

[0018] The following is in conjunction with the appendix Figures 1-4 The specific embodiments of the present invention will be described in detail so that those skilled in the art can fully understand the technical solution of the present invention and implement it.

[0019] I. Overall Structural Composition The silo fabric placement device and control system of this embodiment includes two parts: a silo fabric placement device and a silo fabric placement control system. The two parts achieve data interaction and command transmission through wired (Ethernet) or wireless (5G / Wi-Fi) communication. The whole system is adapted to cylindrical silos with a diameter of 8-15m and is suitable for various materials such as grain, coal, compound fertilizer granules, and cement powder.

[0020] (a) Structure of the fabric distribution device for silos The fabric distribution device includes a material flow monitoring module, an execution module, a control module, and an anomaly alarm module. The physical connections between these modules are as follows: The material flow monitoring module includes 8 flow sensors, 4 density sensors, 6 level sensors, and 2 material characteristic sensors. The flow sensors are evenly distributed at the top of the silo (2 at the inlet), the feed end of the distribution chute (2), and the discharge end of the chute (4), employing electromagnetic flow sensors (measurement range 0.5-50 m³ / h, accuracy ±0.5%). The density sensors are installed on both sides of the middle of the chute (symmetrically arranged), employing microwave density sensors (measurement range 0.3-2.5 g / cm³). The level sensors are distributed and fixed along the height of the silo wall (2 at the top, middle, and bottom), employing 3D lidar level gauges (measurement range 0.5-30 m, resolution 1 mm). The level sensors integrate temperature and humidity detection functions (temperature measurement range -20℃-80℃, humidity measurement range 0-100%RH). The material characteristic sensors are installed below the inlet, employing near-infrared spectral sensors to collect material particle size (measurement range 0.1-50 mm) and humidity (measurement range 0-30%RH) data. The signal output terminals of all sensors are electrically connected to the signal input terminals of the control module via shielded cables.

[0021] Execution module: includes fabric chute, vibrating mechanism, drive unit and chute angle adjustment unit.

[0022] Fabric chute: Made of stainless steel, 3-5m in length, with polytetrafluoroethylene anti-stick coating sprayed on the inner wall of the chute, and a tiltable guide plate installed at the discharge end of the chute. Vibration mechanism: includes 6 high-frequency vibration motors (model: YZU-10-2, power 1.1kW, vibration frequency 50Hz, amplitude 0.5-2mm adjustable), evenly distributed in the lower conical section of the silo (one motor is arranged at every 60° interval), and each vibration motor corresponds to an independent drive controller; Drive unit: includes chute drive motor (model: stepper motor 57HS22, torque 2.2N・m) and vibratory drive module (using PWM pulse width modulation module), which are electrically connected to chute angle adjustment unit and vibratory motor respectively; The chute angle adjustment unit includes a gear transmission mechanism and an angle sensor (measuring range 0-90°, accuracy ±0.1°). The gear transmission mechanism is fixedly connected to the rotating shaft of the fabric chute, and the signal output terminal of the angle sensor is connected to the control module.

[0023] Control module: It adopts a PLC controller (model: S7-1200, CPU1214C), which integrates an analog input module (SM1231), a digital output module (SM1222) and a communication module (supporting Profinet / Modbus protocol). It is installed in a control box outside the silo. The control box has dustproof and waterproof functions (IP65 protection level).

[0024] Abnormal alarm module: includes an audible and visual alarm (model: LTE-1101J, alarm volume ≥85dB) and indicator lights (red, yellow, and green), which are installed on the top of the control box and electrically connected to the digital output terminal of the control module.

[0025] (II) Composition of the fabric control system for silos The control system includes a data acquisition unit, a data processing unit, an analysis and prediction unit, a parameter optimization unit, an execution control unit, a storage unit, and a remote interaction unit. These units interact with each other via an industrial bus or network. Data acquisition unit: A data acquisition card (model: NI9205) is used to communicate with the signal output terminals of each sensor in the material flow monitoring module. It is used to collect flow status data and material characteristic data in real time. The sampling frequency is set to 10Hz and the acquired data format is converted to JSON format.

[0026] The data processing unit is built on an ARM Cortex-A9 processor (1GHz) and incorporates a built-in filtering algorithm (using Kalman filtering) to remove noise from the collected raw data, generating purified flow data sets. Abnormal data points are marked using preset thresholds. The threshold setting rules are: lower limit of flow threshold = effective silo volume × 0.01 / feeding cycle (h); upper limit = maximum output flow rate of the silo feed pump × 0.9 (e.g., for a silo with a diameter of 10m and an effective volume of 500m³, a feeding cycle of 12h, and a maximum output flow rate of 40m³ / h for the feed pump, the lower limit of the flow threshold for grain materials = 500 × 0.01 / 12 ≈ 0.42m³ / h, rounded to 0.5m³ / h; the upper limit = 40 × 0.9 = 36m³ / h, rounded to 35m³ / h). Combined with material characteristic information (particle size, humidity), the abnormal data is categorized into three types: "particle size-related abnormalities," "humidity-related abnormalities," and "flow velocity abnormalities," resulting in a categorized subset of abnormal data.

[0027] Analysis and prediction unit: including support vector machine (SVM) model module and neural network model module, both deployed on edge computing gateway (model: EC200).

[0028] The SVM model module employs the RBF kernel function. The training data features are 8-dimensional (feed flow rate, chute outlet flow rate, material density, material level, particle size, humidity, flow velocity deviation, density deviation), and are categorized into 4 types (normal / mild blockage / heavy blockage / uneven discharge). Cross-validation is used to optimize the RBF kernel function with a penalty coefficient C=10 and gamma=0.1. The input features are a subset of classified outlier data, and the output is the material flow state category. The model training set consists of 100,000 sets of historical data (covering different materials and operating conditions), achieving a classification accuracy ≥95%. Neural network model module: A BP neural network (5 neurons in the input layer, 20 neurons in the hidden layer, and 1 neuron in the output layer) is used. The training data input is 5-dimensional (particle size deviation, humidity deviation, flow velocity deviation, density deviation, and historical fluctuation amplitude), and the output is 1-dimensional (predicted fluctuation amplitude for the next 5 minutes). The Adam optimizer is used with a learning rate of 0.001, 500 iterations, and the loss function is mean squared error (MSE). The model is pre-trained based on historical flow state data, vibration parameters, and discharge effect data, with a mean squared error ≤ 0.02.

[0029] The parameter optimization unit is built on an industrial computer (CPU i5-10400, 8GB memory) and incorporates a linear regression model. First, it calculates the real-time discharge fluctuation amplitude (formula: fluctuation amplitude = (maximum discharge flow rate - minimum discharge flow rate) / average discharge flow rate). When the fluctuation amplitude exceeds a preset threshold (0.05), it extracts the core elements of vibration parameters: frequency, amplitude, and timing. The linear regression model calculates the contribution coefficient of each element to the fluctuation amplitude (e.g., frequency contribution coefficient 0.35, amplitude contribution coefficient 0.42, timing contribution coefficient 0.23). Based on these contribution coefficients, it generates five candidate vibration parameter combinations (e.g., combination 1: frequency 50Hz, amplitude 1.2mm, timing interval 3s; combination 2: frequency 55Hz, amplitude 1.0mm, timing interval 2.5s, etc.). Simulation is then used to select the optimal combination with the lowest discharge fluctuation amplitude.

[0030] The execution control unit uses a relay module (model: OMRONG2R-1) to convert the optimized vibration parameter combination into control signals (voltage 0-10V, current 4-20mA), which are sent to the drive unit of the execution module. Simultaneously, it generates chute angle adjustment commands (based on bin level data collected by level sensors, e.g., 30° for low position, 45° for middle position, and 60° for high position), controlling the chute angle adjustment unit. If external equipment (such as ventilation equipment) needs to be linked, it communicates with the ventilation equipment controller via the Modbus-RTU protocol. The control commands are digital signals (high level 1 = start, low level 0 = stop). The triggering logic is: when the humidity inside the bin exceeds 25%, a high-level signal is output to start the ventilation equipment, and a low-level signal is output to stop it when the humidity drops below 20%.

[0031] Storage unit: SD card (capacity 64GB) + cloud storage (Alibaba Cloud OSS) is used. Local storage is used for real-time operation data (retained for 3 months) and cloud storage is used for historical data (retained long-term). The stored content includes flow status data, vibration parameter adjustment records, discharge effect data and material characteristics-closing parameter matching library (e.g., for grain materials: when the humidity is 10-15%, the optimal vibration frequency is 45-50Hz and the amplitude is 0.8-1.2mm).

[0032] The remote interaction unit includes a touchscreen (model: Weintek MT8102iE, 10.1-inch) and a remote monitoring platform (developed based on the Python Flask framework). The touchscreen is mounted on the front of the control box for local parameter settings and status viewing. The remote monitoring platform supports PC (browser access) and mobile (APP access), enabling visual monitoring (real-time display of silo location, material flow status, and vibration parameters), remote parameter adjustment (hierarchical access control), and abnormal alarm push notifications (SMS + APP notifications). It supports centralized management of up to 10 silos.

[0033] II. Specific Work Process The workflow of the silo fabric distribution device and control system in this embodiment follows a closed-loop logic of "monitoring-processing-prediction-optimization-execution-feedback", and the specific steps are as follows: Step 1: System Initialization After the system is started, the control module automatically performs self-checks on each sensor and actuator (such as whether the sensor signals are normal and whether the vibrating motor can start normally). If the self-check passes, the green indicator light on the remote interaction unit will light up, and the storage unit will load the preset parameters of the corresponding material (such as the flow threshold of compound fertilizer granules 1-35m³ / h, the initial value of vibration parameters: frequency 48Hz, amplitude 1.0mm, and timing interval 3s). If the self-check fails, the red indicator light will light up and trigger an audible and visual alarm, while the fault location will be displayed on the touch screen and the remote platform (such as "Flow sensor No. 3 is faulty").

[0034] Step 2: Data Acquisition and Preprocessing Material enters the distribution chute through the feed inlet at the top of the silo. The material flow monitoring module's sensors collect data at a frequency of 10Hz: a flow sensor collects the material flow rate at the feed inlet and chute inlet / outlet; a density sensor collects the material density in the chute; a level sensor collects the material level at different heights within the silo, as well as the temperature and humidity; and a material characteristic sensor collects the material particle size and humidity. The data acquisition unit transmits the raw data to the data processing unit, where a Kalman filter algorithm removes environmental interference (such as noise caused by vibration and dust), generating purified flow data. Data points exceeding the specified range are then flagged using preset thresholds (e.g., flow rates below 1 m³ / h or above 35 m³ / h), and categorized into abnormal data subsets such as "particle size-related anomalies" (e.g., particle sizes exceeding the preset range of 0.5-20 mm).

[0035] Step 3: Status Analysis and Trend Forecasting The analysis and prediction unit receives a subset of classified abnormal data from the data processing unit. This subset includes not only data points marked as abnormal, but also contextual information such as their corresponding material characteristics (particle size, humidity) and the location of occurrence (sensor ID).

[0036] 3.1 State Classification of Support Vector Machine (SVM) Model Feature Vector Construction: The input to the SVM model is a standardized 8-dimensional feature vector. This vector is derived from real-time sensor data through computation, and its specific construction method is as follows: Feed flow rate ratio: (current feed inlet flow rate / design rated flow rate), used to characterize the feed load.

[0037] Chute outlet flow uniformity: (maximum outlet flow - minimum outlet flow) / average outlet flow, directly reflects the uniformity of the fabric.

[0038] Density gradient: (density in the middle of the chute - estimated density at the inlet), used to sense whether the material is piled up and compacted in the chute.

[0039] Material level change rate: (the rate at which the material level at a specific point in the silo decreases per unit time), indirectly reflecting the overall smoothness of material discharge.

[0040] Particle size deviation index: (current average particle size - current standard particle size of material) / standard particle size.

[0041] Humidity deviation index: (current humidity value - current material standard humidity) / standard humidity.

[0042] Regional flow rate attenuation rate: For each feeding area, calculate (current flow rate / average flow rate in the previous 5 minutes) to identify the onset of local blockage.

[0043] Comprehensive fluctuation index: the normalized standard deviation of the aforementioned multiple flow and density data within a short time window.

[0044] Model Inference and Classification: The constructed real-time feature vector is input into the trained SVM model. Based on its decision function, the model outputs a label belonging to a predefined category (e.g., "0-Normal", "1-Slight Blockage", "2-Severe Blockage", "3-Uneven Discharge"). For example, when the "Channel Outlet Flow Balance" and "Regional Flow Decay Rate" in the feature vector are significantly high, and the "Density Gradient" is also positive, the model is more likely to classify it as "3-Uneven Discharge" or "1-Slight Blockage".

[0045] Training Data Foundation: This SVM model was trained using over 100,000 sets of historical operating condition data. Each set of data includes the aforementioned 8-dimensional features and a final state label confirmed by human experts or long-term operating results. The training process optimized the parameters of the RBF kernel function (C=10, γ=0.1) through cross-validation to ensure classification accuracy (≥95%) in complex nonlinear feature spaces.

[0046] 3.2 Trend Prediction Using a BP Neural Network Model Input and Output: This BP neural network is a regression model with a 5-dimensional input layer, specifically: Particle size deviation index (same as above).

[0047] Humidity deviation index (same as above).

[0048] Combined flow velocity deviation (the combined deviation of readings from multiple flow sensors).

[0049] Combined density deviation (the combined deviation of readings from multiple density sensors).

[0050] Historical fluctuation range (standard deviation of discharge flow rate over the past 10 minutes).

[0051] The output layer is one-dimensional, representing the predicted fluctuation range of the output flow rate over the next 5 minutes. This is a specific numerical prediction used to quantitatively assess the risk of unstable output without intervention.

[0052] Network Training and Learning: The neural network uses input-output pairs from historical data for supervised learning. For example, a set of training data might use [particle size deviation, humidity deviation, velocity deviation, density deviation, and fluctuation amplitude in the 10 minutes before time t] as input, with the corresponding label being [the actual fluctuation amplitude from time t to t+5 minutes]. The model continuously adjusts the weights of its hidden layers (20 neurons) using the backpropagation algorithm (Adam optimizer, learning rate 0.001) to minimize the mean squared error (MSE) between the predicted and true values. After sufficient training, the model can capture the deep, non-linear correlation between material property deviations, the current state, and future short-term stability.

[0053] Prediction execution: The 5-dimensional input vector calculated at the current moment is fed into a pre-trained neural network to obtain a predicted value for the future fluctuation range. If this value exceeds a threshold (e.g., 0.05), it is marked as "trend anomaly".

[0054] Step 4: Parameter Optimization and Command Generation The number optimization unit receives the output of step 3: state category label and predicted fluctuation amplitude value.

[0055] 4.1 Parameter Contribution Analysis Based on Linear Regression Model Constructing a regression dataset: The system maintains a dynamic "parameter-effect" regression dataset. After each control adjustment, the system records the vibration parameter combination before the adjustment (frequency F, amplitude A, time series T) as the independent variable, and the "discharge fluctuation amplitude" actually calculated over a period of time after the adjustment (e.g., 3 minutes) as the dependent variable Y. This data is accumulated over time to form the dataset {(F_i, A_i, T_i)->Y_i}.

[0056] Online Regression Analysis: When parameter optimization is required (e.g., the current state is "slight blockage" and the predicted fluctuation amplitude is high), the parameter optimization unit uses historical data near the current unstable state as samples to perform a multiple linear regression analysis: Y = β0 + β1*F + β2*A + β3*T + ε. The fitted coefficients β1, β2, and β3 are the contribution coefficients of vibration frequency, amplitude, and timing to the discharge fluctuation amplitude (Y) under the current operating conditions. The larger the absolute value of the coefficient, the greater the weight of the parameter's influence on the current problem.

[0057] Guiding optimization direction: For example, the analysis results show that β2 (amplitude coefficient) is a large positive value, while β1 (frequency coefficient) is a small negative value. This indicates that under the current "slight blockage" condition, increasing the vibration amplitude may significantly exacerbate fluctuations, while appropriately increasing the frequency may help stabilize the discharge. This quantitative analysis provides clear mathematical guidance for subsequent parameter optimization, avoiding blind trial and error.

[0058] 4.2 Candidate Parameter Generation and Screening Candidate combination generation: Based on contribution coefficient analysis, the system performs a biased search near the "current parameter" point, primarily in the direction inversely proportional to the contribution coefficient. Continuing with the previous example, the system will generate more candidate combinations that "slightly increase the frequency, slightly decrease or maintain the amplitude" (e.g., combination 1: F+5Hz, A-0.1mm, T unchanged; combination 2: F+3Hz, A unchanged, T-0.5s, etc.), rather than uniformly and randomly generating them.

[0059] Simulation screening: Candidate combinations are input into a simplified material flow simulation model (this model is built based on the Discrete Element Method (DEM) or empirical formulas and stored in the system). The system quickly simulates the expected discharge state after executing the parameters and calculates the simulated fluctuation range. At the same time, the system also searches for successful parameters under similar historical conditions in the "Material Properties - Distribution Parameter Matching Library".

[0060] Determining the optimal combination: Considering factors such as comprehensive simulation results, historical matching degree, and energy consumption for parameter adjustment, select the candidate combination with the highest comprehensive score, which is determined as the "optimized vibration parameter combination". For example, combination 2 was ultimately selected because it has low simulation fluctuations, similar successful historical cases, and the increase in energy consumption is within an acceptable range.

[0061] 4.3 Final synthesis of control commands The parameter optimization unit packages the final determined "vibration parameter combination" with the "chute angle adjustment value" (e.g., from 45° to 50°) based on the material level analysis, forming a structured control instruction, which is then sent to the execution control unit. This instruction explicitly includes the new frequency, amplitude, action sequence, and target chute angle for each vibrating motor.

[0062] Step 5: Execution and Feedback The execution control unit transmits the optimized vibration parameter combination and chute angle adjustment command to the control module. The control module controls the vibratory motor to operate according to the new parameters (e.g., adjusting the frequency to 52Hz and the amplitude to 1.3mm) through the drive unit. The chute angle adjustment unit drives the gear transmission mechanism to adjust the fabric chute to 55°, and the angle sensor provides real-time feedback on the adjusted angle to ensure accuracy. If a ventilation equipment start command has been sent, the humidity change in the silo is monitored simultaneously. During execution, each sensor continuously collects data, forming a feedback data stream: if the feedback data shows that the blockage is cleared, the discharge is uniform (the flow rate difference between the discharge ends of different chutes is ≤2m³ / h), and the humidity drops below 20%, the system maintains the current parameters and shuts down the ventilation equipment; if there is no improvement, the parameter optimization unit regenerates candidate combinations until a stable state is reached.

[0063] Step 6: Anomaly Alarm and Remote Control If the material flow status is "severe blockage" and persists for 5 minutes without resolution, or if the sensor data shows that the parameter adjustment exceeds the preset range (e.g., the vibration amplitude exceeds the 2mm upper limit), the yellow indicator light of the abnormal alarm module will illuminate, triggering an audible and visual alarm. Simultaneously, the remote interaction unit will send SMS and APP notifications to the designated management personnel, marking the abnormal location and cause (e.g., "severe blockage in area 2 of the silo cone section") on the remote platform. Management personnel can view real-time data through the remote platform, manually adjust parameters (e.g., increase the vibration amplitude to 1.8mm), or issue a stop command for remote intervention.

[0064] Step 7: Data Storage and Update The storage unit records flow status data, vibration parameter adjustment records, material discharge effect data, and anomaly handling results in real time throughout the entire process. At the same time, it regularly updates the material characteristic-clothing parameter matching library (such as adding the optimal parameter combination when the humidity is 18%), providing data support for the subsequent cloth distribution control of similar materials.

[0065] III. Adaptation Instructions for Different Working Conditions Multi-material mixing conditions: When the silo needs to process two or more materials simultaneously (e.g., material A with a particle size of 5-10mm and a moisture content of 12%; material B with a particle size of 10-15mm and a moisture content of 15%), the material characteristic sensors collect the characteristic data of each material. The data processing unit classifies abnormal data according to material type, and the analysis and prediction unit adjusts the model parameters according to the flow characteristics of different materials (the SVM model label weights are adapted to the proportion of the mixed materials, and the input features of the BP neural network are increased to include the material proportion dimension). The parameter optimization unit generates a combination of vibration parameters adapted to the mixed materials (e.g., frequency 50Hz, amplitude 1.1mm, and time interval 2.8s) to ensure that the mixing ratio deviation of the discharged material is ≤0.5%.

[0066] High temperature and high humidity conditions: When the temperature and humidity sensor in the silo detects that the temperature exceeds 35℃ and the humidity exceeds 25%, the data processing unit marks "humidity-related anomaly". The analysis and prediction unit predicts that the discharge fluctuation will increase. The parameter optimization unit automatically reduces the vibration frequency (e.g., from 50Hz to 45Hz) and increases the amplitude (e.g., from 1.0mm to 1.4mm). At the same time, the execution control unit starts the silo ventilation equipment through the Modbus-RTU protocol to avoid material blockage caused by agglomeration.

[0067] High flow rate operation: When the flow sensor detects that the feed flow rate exceeds 30 m³ / h (close to the upper limit of 35 m³ / h), the chute angle adjustment unit automatically adjusts the chute angle to 60° to increase the discharge channel; the rapping mechanism increases the rapping frequency according to the optimized parameters (such as 55 Hz) to ensure rapid material flow and avoid material accumulation and blockage.

[0068] This specific embodiment discloses in detail the structure, connection relationship and working process of the silo feeding device and control system, covering the core technical features and typical working condition adaptation. Those skilled in the art can make equivalent substitutions to some structures based on the above description without departing from the concept of the present invention (such as replacing the PLC controller with other models with equivalent performance), all of which fall within the protection scope of the present invention.

Claims

1. A silo fabric distribution device, characterized in that: include: The material flow monitoring module includes multiple sensors distributed in multiple locations within the silo, used to collect flow state data and material characteristic data of the material within the silo. The flow state data includes material flow velocity and distribution data, and the material characteristic data includes material particle size and moisture content. The execution module includes a fabric chute, a chute angle adjustment unit for driving the fabric chute, and a vibration mechanism acting on the silo. The vibration mechanism is equipped with independently adjustable vibration parameters, including vibration frequency, amplitude, and action sequence. The control module is communicatively connected to the material flow monitoring module, the chute angle adjustment unit, and the vibrating mechanism, respectively. It is used to receive and process the flow state data and material characteristic data, and generate control commands based on the processing results to synchronously adjust the tilt angle of the fabric chute and the vibration parameters of the vibrating mechanism.

2. The silo fabric distribution device according to claim 1, characterized in that: The sensors include a flow sensor, a density sensor, a level sensor, and a near-infrared spectral sensor; the flow sensor is arranged at the silo inlet, the inlet end and the outlet end of the fabric chute; the density sensor is symmetrically arranged on both sides of the middle of the fabric chute; the level sensor is distributed on the silo wall; and the near-infrared spectral sensor is arranged below the inlet.

3. The silo fabric distribution device according to claim 1, characterized in that: The vibration mechanism includes multiple vibration motors, which are circumferentially distributed to different areas of the lower conical section of the silo. The vibration parameters of each vibration motor can be independently controlled.

4. The silo fabric distribution device according to claim 1, characterized in that: It also includes an anomaly alarm module, which is communicatively connected to the control module and is used to trigger audible, visual and / or remote alarm signals when the material flow monitoring module detects a flow anomaly or when the parameters in the control command exceed a preset safety range.

5. A fabric control system for silos, characterized in that, The control system of the silo fabric distribution device according to any one of claims 1-4 includes: The data acquisition unit is used to collect the flow status data and material characteristic data from the material flow monitoring module in real time; The data processing unit is used to filter and denoise the collected data, and to mark and classify abnormal data based on preset material characteristic thresholds, generating a subset of classified abnormal data. The analysis and prediction unit includes a pre-trained support vector machine model and a neural network model; the support vector machine model is used to classify and identify the material flow state based on the abnormal data subset; the neural network model is used to predict the material discharge continuity trend in future periods based on historical data and the current state. The parameter optimization unit is used to calculate the contribution of each element in the vibration parameters to the discharge stability through a regression analysis model based on the classification and identification results and the discharge continuity trend, and to generate an optimized combination of vibration parameters and an adjustment command for the angle of the material chute. The execution control unit is used to send the optimized vibration parameter combination and angle adjustment command to the execution module of the silo fabric device to drive its operation.

6. The silo fabric distribution control system according to claim 5, characterized in that: The support vector machine model's classification categories include at least normal state, blockage state, and uneven discharge state; the neural network model is a BP neural network model trained based on historical flow state data, vibration parameters, and actual discharge effect data.

7. The silo fabric distribution control system according to claim 5, characterized in that: The parameter optimization unit is further configured to: calculate the real-time output fluctuation amplitude, and determine that the current state is an unstable state when the fluctuation amplitude exceeds a set threshold; The contribution coefficients of vibration frequency, amplitude, and timing to the fluctuation under the unstable state are analyzed based on the regression analysis model. Based on the contribution coefficients, multiple sets of candidate vibration parameter combinations are generated, and the combination that minimizes the predicted fluctuation amplitude is selected as the optimized vibration parameter combination through simulation or historical data matching.

8. The silo fabric distribution control system according to claim 5, characterized in that: It also includes a storage unit for storing historical operating data, parameter adjustment records, and a matching knowledge base containing the correspondence between different material properties and optimal fabric parameters.

9. The silo fabric distribution control system according to claim 5, characterized in that: It also includes a remote interaction unit, which provides a visual monitoring interface, supports remote status monitoring, parameter setting, centralized management and control of the fabric device, and receiving alarm information pushed by the abnormal alarm module.

Images