Horizontal agitator filling body material slurry viscosity real-time detection system

Abstract

The application discloses a horizontal mixer filling body material slurry viscosity real-time detection system, belongs to the field of testing material flow characteristics by means of physical properties, and the system comprises a multi-source sensor array, a signal acquisition and viscosity conversion, a weighted data fusion and a monitoring and early warning module. The multi-source sensor array is arranged with at least three groups of sensor nodes along the axial direction of the bottom of a stirring tank, and synchronously acquires slurry torque and temperature signals. After the signals are converted from analog to digital, viscosity conversion and temperature drift compensation, local viscosity signals are output. The weighted data fusion module generates global viscosity signals through consistency inspection and dynamic weighted calculation. The monitoring module completes data display, storage and sensor fault early warning. The system solves the problem that traditional single-point detection cannot obtain real-time overall viscosity of the slurry.

Description

Technical Field

[0001] This invention relates to the field of testing the flow characteristics of materials by means of physical properties, and specifically to a real-time detection system for the viscosity of filling slurry in a horizontal mixer. Background Technology

[0002] Backfill slurry is a core material in mine backfilling. Its viscosity directly determines the flow resistance during pipeline transportation and the self-leveling performance after being filled into the goaf. Improper viscosity control during mixing can easily lead to pipe blockage or uneven backfill strength. Currently, the viscosity testing of backfill slurry in mining operations mainly relies on manual, timed sampling and offline analysis using rotational viscometers or rheometers in laboratories. This method has a delay of tens of minutes between sampling and obtaining results, failing to reflect the true state of the slurry within the mixer in real time. Furthermore, sampling points are usually limited to the vicinity of the mixing tank outlet, ignoring the significant concentration and rheological gradients within the axial space of the horizontal mixer. In recent years, some studies have attempted to install online viscometers at specific locations in the mixing tank. However, these solutions often employ single-point measurement, using the reading from a single sensor at a fixed point to represent the overall slurry viscosity. For horizontal mixers with a large length-to-diameter ratio, the axial movement and mixing of materials require a certain amount of time, making single-point measurement unable to capture the spatial distribution characteristics of viscosity. Furthermore, filling slurries are non-Newtonian fluids with yield stress, and their viscosity exhibits a non-linear relationship with shear rate. Existing detection methods often assume a simple linear correspondence between torque and viscosity, neglecting the conversion errors caused by non-Newtonian rheological properties. Moreover, during stirring, the slurry temperature can fluctuate by tens of degrees Celsius due to frictional heat generation and ambient temperature changes, significantly affecting the slurry's flow characteristics. Current detection schemes generally lack temperature compensation mechanisms. Regarding data utilization, existing technologies typically use arithmetic averaging or simple threshold judgments for processing multi-point measurement data, lacking a unified framework for dynamic evaluation of sensor node data quality and data fusion. Therefore, there is an urgent need in industrial settings for an online detection system capable of real-time, continuous acquisition of the global viscosity of filling slurries, possessing spatial representativeness and temperature self-compensation capabilities. Summary of the Invention

[0003] In view of the shortcomings of the prior art, the purpose of this invention is to provide a real-time viscosity detection system for the filling slurry of a horizontal mixer, which solves the problem that existing single-point detection methods cannot obtain a representative real-time overall viscosity of the filling slurry in a horizontal mixer. This invention synchronously collects local signals by axially arranging multiple sets of integrated torque and temperature sensors at the bottom of the horizontal mixer's mixing tank. A pre-set viscosity-torque conversion model based on the constitutive equation of non-Newtonian fluids is used to convert the torque signal into the original viscosity. The temperature-viscosity compensation relationship is used to eliminate the influence of temperature drift and generate a compensated local viscosity. Then, a weighted data fusion algorithm is used to dynamically verify the consistency of data from each node and assign weights, calculating and outputting a global viscosity signal characterizing the overall state of the slurry.

[0004] This invention provides a real-time viscosity detection system for filling slurry in a horizontal mixer, comprising:

[0005] The multi-source sensor array module has at least three sets of viscosity sensor nodes arranged axially at intervals along the bottom of the mixing tank of the horizontal mixer. Each set of nodes integrates a torque detection unit and a temperature sensing unit to synchronously collect local torque simulation signals and local temperature simulation signals of the slurry.

[0006] The signal acquisition and viscosity conversion module receives the above signal and performs analog-to-digital conversion. It uses a preset viscosity-torque conversion model based on the rheological properties of non-Newtonian fluids to convert the torque digital signal into the original local viscosity signal. Then, based on the temperature-viscosity compensation relationship, it uses the temperature digital signal to compensate for the temperature drift of the original local viscosity signal and generates the compensated local viscosity signal.

[0007] The weighted data fusion module receives the compensated local viscosity signals of all sensor nodes within the same sampling period, performs tests on them, dynamically calculates the weighting coefficients of each node, and generates a global viscosity signal that characterizes the overall state of the slurry in the mixing tank through weighted fusion.

[0008] The monitoring and early warning module receives global viscosity signals and node anomaly flag signals, displays the global viscosity signals in real time and stores them historically, and outputs a sensor fault early warning signal when the node anomaly flag signal indicates that the sensor node data is abnormal.

[0009] In one embodiment of the present invention, each group of viscosity sensor nodes in the multi-source sensor array module further integrates an embedded microprocessor and a node communication unit. The embedded microprocessor is connected to the torque detection unit and the temperature sensing unit, and is used to drive the torque detection unit and the temperature sensing unit to synchronously perform signal acquisition at a preset sampling frequency, and convert the acquired local torque analog signal and local temperature analog signal into local torque digital signal and local temperature digital signal. The node communication unit is connected to the embedded microprocessor, responds to data requests through an industrial serial communication bus, and encapsulates the local torque digital signal and local temperature digital signal into a data packet and transmits it to the signal acquisition and viscosity conversion module. Each group of viscosity sensor nodes shares the same clock synchronization pulse to ensure the consistency of sampling time.

[0010] In one embodiment of the present invention, the viscosity-torque conversion model based on the non-Newtonian fluid rheological properties pre-set in the signal acquisition and viscosity conversion module is established in the following way: Based on the solid mass concentration, fine particle content ratio, and cementitious material dosage of the filling slurry, the yield stress parameter, consistency coefficient parameter, and rheological index parameter are determined using a pre-calibrated regression function. Then, based on the geometric dimensions of the stirring blades and the real-time stirring speed, the stirring speed is mapped to the average shear rate borne by the slurry. Combining the yield stress parameter, consistency coefficient parameter, and rheological index parameter, the apparent viscosity of the slurry under the current conditions is calculated according to the constitutive relationship of non-Newtonian fluids. A correspondence is established between this apparent viscosity and the local torque measured by the torque detection unit, so that the signal acquisition and viscosity conversion module can directly convert the original local viscosity signal from the real-time input local torque digital signal. The calculation formula for the apparent viscosity is as follows:

[0011] ;

[0012] in, The apparent viscosity of the filling slurry, For yield stress parameters, Where is the average shear rate, K is the consistency coefficient parameter, n is the rheological index parameter, and C is the solids mass concentration. This refers to the percentage of fine particles. The real-time stirring speed is given by r, where r is the geometric radius of the stirring blades. This refers to the dosage of cementitious materials.

[0013] In one embodiment of the present invention, the preset temperature-viscosity compensation relationship in the signal acquisition and viscosity conversion module is established based on an Arrhenius-type temperature-viscosity correlation. This relationship describes the quantitative impact of the difference between the local slurry temperature and the preset reference temperature acquired at any sampling time on the slurry viscosity value. The signal acquisition and viscosity conversion module uses the temperature digital signals synchronously uploaded by each sensor node to determine the current local temperature value. The original local viscosity signal is multiplied by a temperature compensation factor to generate a compensated local viscosity signal. The temperature compensation factor is jointly determined by the flow activation energy of the filling slurry, the universal gas constant, the preset reference temperature, and the current local temperature value. This ensures that the compensated local viscosity signals output by the system under different slurry temperature conditions are all reduced to the equivalent viscosity value under the preset reference temperature condition. The calculation formula for the temperature compensation factor is as follows:

[0014] ;

[0015] in, For temperature compensation factor, The value is the activation energy of the filling slurry flow, and R is the universal gas constant. For preset reference temperature, Here is the current local temperature value, n is the rheological index parameter, and K is the consistency coefficient parameter.

[0016] In one embodiment of the present invention, the weighted data fusion module performs consistency checks on the compensated local viscosity signals of all sensor nodes within the same sampling period as follows: calculate the mean of all compensated local viscosity signals, and determine whether the deviation between the compensated local viscosity signal corresponding to each sensor node and the mean exceeds a preset deviation threshold. If the deviation of the compensated local viscosity signal of a certain sensor node exceeds the preset deviation threshold, then the sensor node is marked as a suspicious node in the current sampling period, the compensated local viscosity signal corresponding to the suspicious node is excluded in the current weighted fusion calculation, and the node abnormality flag signal of the suspicious node is set to abnormal and transmitted to the monitoring and early warning module to trigger the output of the sensor fault early warning signal.

[0017] In one embodiment of the present invention, the weighted data fusion module dynamically calculates the weighting coefficients of the compensated local viscosity signals of valid sensor nodes that have passed the consistency check. Specifically, it maintains historical data statistics records for each sensor node, including the data variance and fluctuation amplitude of the node in multiple consecutive sampling periods. If the data variance of a sensor node is lower than a preset variance threshold in a preset number of consecutive sampling periods, it indicates that the slurry at the location of the sensor node tends to be uniform, and the weighting coefficient corresponding to the node is increased. If the data fluctuation amplitude of a sensor node exceeds a preset fluctuation threshold, the weighting coefficient corresponding to the node is decreased. After the weighting coefficients are normalized, the weighted data fusion module performs a weighted summation of the compensated local viscosity signals of each valid sensor node that has passed the consistency check with their respective normalized weighting coefficients to generate the global viscosity signal for the current sampling period. The calculation formula for the global viscosity signal is as follows:

[0018] ;

[0019] in, This is the global viscosity signal. The local viscosity signal after compensation at the i-th node. The normalized weighting coefficient for the i-th node is... Let be the variance of the historical data of the i-th node. This is the weighting coefficient fluctuation correction amount.

[0020] In one embodiment of the present invention, the weighted data fusion module is further configured with a ring-shaped buffer storage area. The ring-shaped buffer storage area is used to store the global viscosity signal generated by each weighted fusion calculation and the compensated local viscosity signal, weighting coefficient and node anomaly flag signal corresponding to each sensor node in the sampling period in the order of sampling period. The storage capacity of the ring-shaped buffer storage area is sufficient to accommodate a preset number of historical data records generated by the system during continuous operation. When the weighted data fusion module receives a data backtracking request from the monitoring and early warning module, it extracts historical data records within a specified time range from the ring-shaped buffer storage area and uploads them to the monitoring and early warning module for viscosity change trend analysis and post-event source tracing of abnormal operating conditions.

[0021] In one embodiment of the present invention, after receiving the global viscosity signal and node anomaly flag signal transmitted in real time by the weighted data fusion module, the monitoring and early warning module presents the time evolution process of the global viscosity signal on the operation interface in the form of numerical display and trend curve, and uses a hierarchical color coding method to visually distinguish the global viscosity signal in different numerical ranges. When the node anomaly flag signal indicates that there is a data anomaly in at least one sensor node, the monitoring and early warning module highlights the identifier of the abnormal sensor node and its corresponding local location information on the operation interface, and generates a sensor fault early warning signal containing the abnormal node identifier, the time of the anomaly, and the viscosity deviation amplitude, and prompts the operator to perform sensor maintenance and inspection in the form of an audible and visual alarm.

[0022] In one embodiment of the present invention, the monitoring and early warning module further continuously judges the abnormal sensor node data indicated by the node abnormality flag signal. If the node abnormality flag signal of the same sensor node is continuously in an abnormal state within a preset number of sampling periods, the monitoring and early warning module marks the sensor node as a fault node, outputs a sensor fault confirmation signal to the operator through the operation interface, and sends a node disable command to the weighted data fusion module through the communication interface in the subsequent viscosity global calculation process. The command instructs the weighted data fusion module not to introduce the compensated local viscosity signal corresponding to the fault node into the weighted fusion calculation from the next sampling period, and to re-normalize the weighting coefficients of the remaining valid sensor nodes to maintain the continuous output of the global viscosity signal.

[0023] In one embodiment of the present invention, the monitoring and early warning module also receives externally input slurry quality parameter signals. The slurry quality parameter signals include at least the slurry solid mass concentration and the proportion of fine particles. The monitoring and early warning module aligns the slurry quality parameter signals with the global viscosity signal at the corresponding time according to the timestamp and stores them together. When the global viscosity signal exceeds the preset viscosity allowable range, the monitoring and early warning module automatically compares the current slurry quality parameter signals with the historically stored slurry quality parameter signals and performs deviation analysis. It generates control prompt signals that suggest adjusting the slurry solid mass concentration and the proportion of fine particles and displays them to the operator on the operation interface to assist in the mixing quality control of the filling slurry.

[0024] Beneficial effects: The present invention provides a real-time viscosity detection system for filling slurry in a horizontal mixer. By axially arranging multiple sets of integrated torque and temperature sensors at the bottom of the mixing tank of the horizontal mixer, local signals are collected synchronously. The torque signal is converted into the original viscosity using a pre-set viscosity-torque conversion model based on the constitutive equation of non-Newtonian fluids. The temperature drift effect is eliminated according to the temperature-viscosity compensation relationship to generate compensated local viscosity. Then, the consistency of data from each node is dynamically checked and weights are assigned through a weighted data fusion algorithm. Finally, a global viscosity signal characterizing the overall state of the slurry is calculated and output. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a system architecture diagram of a real-time viscosity detection system for filling slurry in a horizontal mixer.

[0027] Figure 2 This is a schematic diagram of the overall workflow;

[0028] Figure 3 This is a flowchart illustrating the signal acquisition and viscosity conversion module.

[0029] Figure 4 This is a flowchart illustrating the weighted data fusion module and the monitoring and early warning module. Detailed Implementation

[0030] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0031] Please see Figures 1-4This invention provides a real-time viscosity detection system for filling slurry in a horizontal mixer, comprising a multi-source sensor array module. At least three sets of viscosity sensor nodes are axially spaced along the bottom of the mixing tank of the horizontal mixer. Each set of nodes integrates a torque detection unit and a temperature sensing unit for synchronously acquiring simulated local torque and temperature signals of the slurry. A signal acquisition and viscosity conversion module receives the signals and performs analog-to-digital conversion. Using a pre-set viscosity-torque conversion model based on the rheological properties of non-Newtonian fluids, the digital torque signal is converted into the original local viscosity signal, and then calculated based on temperature-viscosity compensation. The system uses digital temperature signals to compensate for temperature drift in the original local viscosity signal, generating a compensated local viscosity signal. The weighted data fusion module receives the compensated local viscosity signals from all sensor nodes within the same sampling period, performs verification on them, and dynamically calculates the weighting coefficients for each node. Through weighted fusion, it generates a global viscosity signal that characterizes the overall state of the slurry in the mixing tank. The monitoring and early warning module receives the global viscosity signal and node anomaly flag signals, displays the global viscosity signal in real time and stores it historically. When the node anomaly flag signal indicates that the sensor node data is abnormal, it outputs a sensor fault early warning signal.

[0032] like Figure 1 As shown, the detection system of the present invention consists of four parts working together: a multi-source sensor array module deployed at the bottom of a horizontal mixer, a signal acquisition and viscosity conversion module responsible for signal conversion and compensation, a weighted data fusion module that performs data fusion processing, and a monitoring and early warning module for operators. This forms a complete technical link from physical signal perception to global viscosity output and then to abnormal state early warning.

[0033] like Figure 2As shown, the system startup initialization module is the starting point of the entire detection process. It completes basic preparatory work such as hardware self-testing, parameter presetting, clock calibration, and communication link establishment, ensuring that each module is in standard working condition and providing a stable operating environment for subsequent signal acquisition and data processing. The multi-source sensor array module, as the core unit of data acquisition, evenly distributes multiple sets of sensor nodes along the axial direction of the bottom of the horizontal mixer's mixing tank, simultaneously acquiring simulated torque and temperature signals of the slurry. This overcomes the limitations of single-point detection and comprehensively captures the slurry state information at different locations within the mixing tank. The signal acquisition and viscosity conversion module receives the acquired signals, performs analog-to-digital conversion, viscosity conversion, and temperature drift compensation, transforming the original physical signals into standardized local viscosity data and eliminating detection errors caused by temperature fluctuations and non-Newtonian fluid characteristics. The weighted data fusion module integrates and processes the standardized data from multiple nodes. Through data verification and dynamic weighted calculation, it eliminates abnormal data interference, generates a global viscosity signal that truly reflects the overall state of the slurry, and simultaneously generates node anomaly flag signals to mark the validity of the data. The node data validity judgment module is a core branch node in the process. It classifies and judges the data status based on abnormal indicator signals. If the data is normal, it enters the real-time display and historical storage stage of the monitoring and early warning module. This module presents the viscosity data change trend through a visual interface and simultaneously completes data archiving and storage, providing a basis for subsequent operational analysis. If the data is abnormal, it directly triggers the sensor fault early warning module, informing the equipment of operational abnormalities through audible and visual prompts and information annotations, assisting maintenance personnel in quickly locating the problem. The continuous detection module constitutes the loop logic of the process. Based on the system operation instructions, it determines whether to continue the detection task. If continuous detection is confirmed, it returns to the multi-source sensor array module to start a new round of signal acquisition. If detection is terminated, it enters the system stop module to complete data saving, equipment power-off, and process closing. The entire process has a clear vertical hierarchy, and the branch and loop logic balances the real-time nature of detection, stability, and timely fault handling.

[0034] A multi-source sensor array module is arranged axially at intervals along the bottom of the horizontal mixer's mixing tank, containing no fewer than three sets of sensor nodes. Each set of sensor nodes structurally integrates two types of sensitive elements: the first is a torque detection unit, used to sense the rotational resistance generated by the filling slurry flowing within the mixing tank on the immersion rotor, and outputs a local torque analog signal corresponding to this resistance; the second is a temperature sensing unit, used to synchronously sense the local temperature of the slurry at the node's location, and output a local temperature analog signal corresponding to this temperature. Multiple sets of nodes are distributed along the mixer's axial direction at different spatial locations, enabling the system to simultaneously acquire the torque response and temperature distribution of the slurry in different areas of the mixing tank (front, middle, and rear sections) within a single sampling cycle, thus providing comprehensive coverage of the spatial gradient information of the slurry viscosity at the physical level. Each sensor node operates independently, with each node triggered for sampling by a unified external synchronization clock, ensuring strict alignment of sampling times across all nodes in the time dimension. This lays the foundation for the comparability and temporal consistency of data from different locations in subsequent fusion calculations.

[0035] The signal acquisition and viscosity conversion module receives local torque and local temperature analog signals from various nodes in the multi-source sensor array module. This module first performs analog-to-digital conversion on the input analog signals, converting each continuously changing analog electrical signal into discrete digital quantities, forming local torque and local temperature digital signals. After obtaining the digital signals, the signal acquisition and viscosity conversion module calls its internally preset viscosity-torque conversion model based on the rheological properties of non-Newtonian fluids to convert the local torque digital signals of each node into the original local viscosity signals. This viscosity-torque conversion model is not a simple linear proportional relationship, but is based on the rheological laws followed by the filling slurry as a non-Newtonian fluid with yield stress. Specifically, the model fully considers the influence of factors such as the variation range of solid mass concentration of the filling slurry, the difference in the proportion of fine particles, and the amount of cementitious materials on macroscopic rheological parameters during the modeling stage. It maps the shear rate distribution within the stirred tank, determined by the blade geometry and stirring speed, to the average shear rate of the slurry, and then calculates the apparent viscosity of the slurry corresponding to the current torque state based on the nonlinear constitutive relationship between shear stress and shear rate in non-Newtonian fluids. Through this model, the signal acquisition and viscosity conversion module can convert physical quantities at the torque sensing level into viscosity values ​​with rheological physical meaning. The formula for calculating apparent viscosity is as follows:

[0036] ;

[0037] This formula integrates the constitutive relationship of non-Newtonian fluids with the core quality parameters of the filling slurry, accurately converting the local torque signal collected by the torque detection unit into the original local viscosity signal. This solves the problem that traditional viscosity detection cannot adapt to the rheological characteristics of non-Newtonian filling slurries, takes into account both linear and nonlinear correlations between parameters, and ensures that the conversion accuracy of the original local viscosity signal can match the real-time stirring conditions of the horizontal mixer. This provides accurate basic data for subsequent temperature compensation and data fusion. The calculation results directly determine the accuracy of the initial viscosity detection of the system and are the core algorithm foundation of the entire real-time viscosity detection system.

[0038] like Figure 3 As shown, this process focuses on the front-end signal acquisition and viscosity conversion stages, using clock synchronization as the foundation and periodic loops as the core to construct a high-precision signal processing system. The clock synchronization pulse trigger module provides a unified time reference for the entire acquisition process, ensuring that all sensor nodes start signal acquisition at the same time, guaranteeing the synchronization and comparability of multi-source data from the source. The sensor node synchronous sampling module receives clock pulse commands, driving each node to perform sampling operations according to a preset frequency, ensuring the continuity and regularity of data acquisition. The torque / temperature signal acquisition module, through integrated sensing units, accurately acquires the local torque and temperature analog signals of the slurry, completely recording the physical state parameters of the slurry, providing raw data support for subsequent conversions. The embedded microprocessor analog-to-digital conversion module converts analog signals into digital signals, improving the anti-interference capability and processing accuracy of signal transmission, adapting to the digital signal requirements of subsequent algorithm calculations. The viscosity-torque model conversion module establishes a conversion relationship based on the rheological properties of non-Newtonian fluids, combining key indicators such as slurry concentration, particle content, and stirring parameters to accurately convert the torque digital signal into the original local viscosity signal, closely matching the fluid characteristics of the filling slurry and improving the accuracy of viscosity detection. The temperature drift compensation calculation module, based on the temperature-viscosity correlation model, uses real-time temperature data to correct the original viscosity signal, eliminating the interference of temperature changes on the viscosity detection results and generating a standardized compensated local viscosity signal. The rheological parameter calibration module provides basic parameter support for viscosity-torque model conversion, ensuring the accuracy of the model conversion through pre-calibrated parameters such as yield stress and consistency coefficient. The temperature compensation model establishment module provides the algorithmic basis for temperature drift compensation calculation, quantifying the correlation between temperature and viscosity based on the Arrhenius correlation to ensure the compensation effect closely matches actual working conditions. The output compensated local viscosity module transmits the processed standard data externally, providing input for subsequent data fusion. The sampling cycle end module is a loop judgment node in the process, determining whether the current sampling cycle is complete. If not, synchronous sampling continues; if completed, it enters the waiting for the next cycle module, maintaining the periodicity and continuity of system sampling. The entire process is tightly integrated vertically, combining parameter support with cyclic sampling to achieve high-precision and high-synchronization processing of the front-end signal.

[0039] After the torque-to-viscosity conversion is completed, the signal acquisition and viscosity conversion module further performs temperature drift compensation on the original local viscosity signal based on its internally preset temperature-viscosity compensation relationship to generate a compensated local viscosity signal. This temperature-viscosity compensation relationship is established based on the physical law that the viscosity of the filling slurry decreases with increasing temperature, using the local temperature digital signals synchronously acquired by each sensor node as the input variable for compensation calculation. Its basic compensation logic is as follows: First, a preset reference temperature is determined as the base temperature for viscosity conversion. Then, the difference between the actual local temperature at any sampling time and the preset reference temperature is quantified as a temperature difference value. Next, the sensitivity of temperature to viscosity is determined based on the specific flow activation energy parameters of the filling slurry and the universal gas constant. Finally, the original local viscosity is multiplied by a temperature compensation factor determined by the above parameters, and the viscosity value under the measured temperature conditions is reduced to the equivalent viscosity value under the reference temperature conditions. After temperature drift compensation, the compensated local viscosity signals output by each sensor node in the same sampling period have eliminated measurement biases introduced by temperature inconsistencies or time fluctuations at different node locations. Data from different nodes are now unified in the temperature dimension, providing the foundation for direct comparison and fusion. The formula for calculating the temperature compensation factor is as follows:

[0040] ;

[0041] in, For temperature compensation factor, The value is the activation energy of the filling slurry flow, and R is the universal gas constant. For preset reference temperature, Here, is the current local temperature value, n is the rheological index parameter, and K is the consistency coefficient parameter. This formula accurately quantifies the impact of the difference between the local temperature of the slurry and the preset reference temperature on the viscosity value. It establishes a nonlinear relationship between temperature and viscosity through the flow activation energy and the gas constant, which conforms to the physical law of viscosity change with temperature in filling slurries. At the same time, a rheological parameter correction compensation factor is introduced to avoid the calculation deviation caused by single temperature compensation. The signal acquisition and viscosity conversion module multiplies the original local viscosity signal with the temperature compensation factor to obtain the compensated local viscosity signal under the preset reference temperature condition. This solves the viscosity detection drift problem caused by slurry temperature fluctuations, enabling the system to output stable and comparable viscosity detection data under different temperature conditions. It provides a standardized input signal for subsequent multi-node data weighted fusion and is a key algorithm link connecting the original viscosity calculation and global data fusion.

[0042] The weighted data fusion module is the core processing unit connecting the front-end array sensing and the back-end global state output. Within the same sampling period, this module receives compensated local viscosity signals generated by all sensor nodes from the signal acquisition and viscosity conversion module, using these as the raw input set for the current fusion calculation. To ensure the reliability of the fused data, the weighted data fusion module first performs a consistency check on the input set. The check is performed by calculating the statistical mean of all compensated local viscosity signals within the current period and using this mean as a benchmark to measure the deviation of each node's value. If the deviation of a node's compensated local viscosity signal from the mean exceeds a preset consistency criterion threshold, it indicates a significant difference between the node's measurement result at the current location and the measurement results of most other nodes. Based on this, the system determines that the node may have a measurement anomaly, excludes it from the fusion calculation for the current period, and generates a corresponding node anomaly flag signal for subsequent early warning. For the remaining valid node data retained after the consistency check, the weighted data fusion module enters the dynamic weight calculation stage. This module maintains a set of historical data statistics for each sensor node in the system, continuously tracking the stability and fluctuation characteristics of the output data of each node in past consecutive sampling periods. When the value of the compensated local viscosity signal output by a sensor node changes little and the variance remains at a low level over multiple consecutive sampling periods, it indicates that the slurry flow state at the node's location is relatively stable, and the reliability of the detection data is high. The weighted data fusion module accordingly increases the weighting coefficient of that node in the current fusion. Conversely, if the data of a node shows frequent and large fluctuations recently, its weighting coefficient is reduced to suppress the impact of this highly uncertain data on the global result. After the weighting coefficients of each effective node are normalized, the weighted data fusion module performs a weighted summation of the compensated local viscosity signal of each effective node with its normalized weighting coefficient to calculate the global viscosity signal that characterizes the overall state of the filling slurry in the mixing tank in the current sampling period. Due to the introduction of weighting coefficients, this global viscosity signal not only integrates local information from multiple spatial locations within the stirred tank, but also adaptively suppresses the influence of suspicious or drastically fluctuating nodes, improving the robustness and representativeness of the data fusion results. The formula for calculating the global viscosity signal is as follows:

[0043] ;

[0044] in, This is the global viscosity signal. The local viscosity signal after compensation at the i-th node. The normalized weighting coefficient for the i-th node is... Let be the variance of the historical data of the i-th node. This is the weighted coefficient fluctuation correction amount. This formula achieves optimized fusion of multi-source sensor data through hierarchical calculation. First, it performs a basic weighted summation on the valid node data that passes the consistency check. Then, it combines data variance for stability correction. Finally, it eliminates the influence of extreme values ​​in the weighted coefficients, ensuring the reliability and accuracy of the global viscosity signal. All parameters in the formula are directly related to the output results of the previous temperature compensation and viscosity conversion algorithms, forming a complete algorithm chain of "original viscosity calculation - temperature compensation correction - dynamic weighted fusion". The weighted data fusion module integrates multi-node data based on this formula, and simultaneously marks abnormal nodes using the consistency check, ensuring that the global viscosity signal can truly reflect the overall viscosity state of the slurry in the mixing tank. This provides accurate data support for real-time display, fault warning, and quality control of the monitoring and early warning module. It is the core algorithm for achieving efficient utilization of multi-source sensor data and improving the system's detection accuracy and robustness.

[0045] The monitoring and early warning module receives the global viscosity signal and node anomaly flag signals output in real time from the weighted data fusion module, and performs three functions: data presentation, historical storage, and anomaly status alerts. Regarding data presentation, the module displays the global viscosity signal on the user interface in both numerical and trend curve formats, allowing operators to directly observe the overall viscosity level of the filling slurry at the current moment and the continuous viscosity change trend throughout the mixing process. For historical storage, the module stores the global viscosity signal of each sampling period in chronological order into a database or storage medium, forming a traceable viscosity history record, providing data support for subsequent batch quality analysis and process parameter optimization. Regarding anomaly status alerts, the module continuously monitors the status changes of the node anomaly flag signals. When the node anomaly flag signal from the weighted data fusion module indicates that at least one sensor node's data is determined to be abnormal, the module prominently displays the installation location and node number of the abnormal node on the user interface, and simultaneously notifies the operator in the form of a sensor fault early warning signal, prompting them to check and maintain the designated sensor node. This mechanism allows operators to promptly obtain abnormal information about the sensor's operating status without having to continuously monitor the raw data from each sensor node.

[0046] like Figure 4As shown, this process revolves around multi-source data fusion and intelligent monitoring and early warning, vertically connecting all stages of data processing, anomaly detection, and source tracing analysis. It incorporates multiple branches and loop logic to enhance the system's intelligence level. The process begins with the local viscosity module after compensation, receiving standardized viscosity data from the front end, completing data reception and temporary storage to ensure data input stability. The consistency verification module performs deviation analysis on multi-node data within the same period, calculating the mean and deviation values ​​to filter out abnormal data deviating from the normal range, ensuring the validity of the data participating in the fusion. The data deviation judgment module is the core branch node, classifying data based on the verification results. If the deviation exceeds the limit, it enters the suspicious node marking and anomaly flag output module, marking abnormal nodes and generating anomaly signals to provide a basis for subsequent early warnings; if the deviation is normal, it directly enters the dynamic weighting coefficient calculation module. The dynamic weighting coefficient calculation module combines the variance and fluctuation range of historical data for each node to dynamically adjust the weight of each node's data. Higher data stability results in greater weight, improving the credibility of the fusion results. The weighted fusion calculation module combines effective node data with corresponding weights, completing the integration processing of multi-source data through weighted calculations. The global viscosity signal generation module outputs final test data that characterizes the overall state of the slurry, providing a core basis for monitoring and control. The real-time display / historical storage module visualizes and archives the global viscosity signal, comprehensively recording system operation data. The continuous anomaly detection module continuously monitors the status of abnormal nodes. If an anomaly persists, it marks the faulty node, disables its data module, locks the faulty node, removes its data, and readjusts the weights of valid nodes to ensure uninterrupted testing. If there are temporary anomalies or the data is normal, it enters the normal testing maintenance module to keep the system running normally. The data backtracking request module constitutes the traceability branch of the process. Upon receiving a backtracking instruction, it enters the circular buffer to read historical data, extracting historical data for a specified time period to assist in operational condition analysis and anomaly tracing. If no backtracking request is received, it returns to the data receiving module to start a new round of processing. The entire process is vertically hierarchical, with branch logic covering anomaly handling, fault locking, and data tracing. The loop logic ensures the continuity of testing, comprehensively achieving the accuracy of data fusion and the reliability of system operation.

[0047] The information flow between the four modules exhibits a bottom-up, unidirectional progressive relationship. The multi-source sensor array module, as the system's physical sensing terminal, converts the mechanical and thermal effects generated during the mixing of the filling slurry into electrical signals suitable for subsequent processing, serving as the data source for the entire system. The signal acquisition and viscosity conversion module receives the raw sensing signals, converts them from analog to digital quantities, and applies rheological modeling and temperature compensation to elevate the raw signals into standardized viscosity data with physical consistency, thus enhancing the signal domain into the information domain. The weighted data fusion module, based on the comparable compensated local viscosity data from each front-end node, uses a dual strategy of statistical verification to eliminate outliers and assigning confidence weights based on historical performance to aggregate multi-point distributed information into a single viscosity output with global representation capabilities, completing the synthesis from multi-source heterogeneous local information to overall state information. The monitoring and early warning module ultimately delivers this global state information to the operators in a visual manner and proactively intervenes to prompt maintenance when sensors malfunction. The entire technical system realizes a complete processing flow for the viscosity of the filling slurry in the horizontal mixer, from spatially distributed synchronous sensing to unified measurement after temperature compensation and then to weighted fusion global characterization, through modular functional division and standardized signal interfaces. This enables the system to continuously output real-time viscosity detection results with spatial representativeness and temporal consistency under the complex working conditions of industrial mixing sites.

[0048] In one embodiment of the present invention, a copper mine backfilling station uses a twin-shaft horizontal mixer to prepare whole tailings paste backfill slurry. The effective length of the mixing tank is 6.5m, the inner diameter of the tank is 1.8m, the diameter of the mixing blades is 1.6m, and the normal operating speed is 45rpm. The designed solids concentration of the slurry is 72%, the cementing material is ordinary Portland cement, and the admixture is 12% of the dry weight of the tailings. The content of -20μm fine particles in the whole tailings used is about 35%. The system has 5 sets of sensor nodes installed at equal intervals along the axial direction at the bottom of the mixing tank, located at 0.5m, 1.8m, 3.1m, 4.4m and 5.7m from the feed end, respectively, and numbered as node S1 to node S5.

[0049] Before system startup, the rheological parameters in the viscosity-torque conversion model were calibrated. Technicians extracted rheological test data from historical backfilling records of the mine's backfilling station, covering 24 mix proportions with total tailings solid mass concentrations ranging from 68% to 74%, fine particle content from 30% to 40%, and cement content from 10% to 14%. A multivariate nonlinear regression method was used to establish regression functions between yield stress, consistency coefficient, and rheological index and these three factors. Regression determined that, under the current batch target mix proportions, the yield stress τ0 was 142 Pa, the consistency coefficient K was 8.7 Pa·sⁿ, and the rheological index n was 0.68. Based on the stirring blade radius of 0.8 m and the stirring speed of 45 rpm, the blade tip linear velocity was calculated to be 3.77 m / s, and the corresponding effective average shear rate was determined to be 12.5 s⁻¹. These parameters were substituted into the Herschel-Balkley constitutive relation to establish a mapping function between torque and apparent viscosity, which was then written into the firmware of the signal acquisition and viscosity conversion module. Regarding temperature compensation parameters, the activation energy Ea of the flow of the whole tailings paste slurry was determined to be 2.85 × 10⁻⁶ through preliminary experiments. 4 The parameters J / mol, the universal gas constant R is 8.314 J / (mol·K), and the preset reference temperature T_ref is set to 25℃. These parameters are also pre-written into the compensation program.

[0050] After the mixer starts, tailings, cement, and thickening water are continuously fed into the mixing tank according to the mix ratio. The materials are gradually mixed under the action of the biaxial blades and propelled towards the discharge end. The system sampling period is set to 1 second. The torque detection unit and temperature sensing unit of the sensor nodes are triggered synchronously by a unified clock pulse at the beginning of each sampling period. During a sampling period after the mixer has entered steady-state operation, the data output by the five sensor nodes are as follows: Node S1: Local torque 0.412 N·m, local temperature 31.2℃; Node S2: Local torque 0.385 N·m, local temperature 30.6℃; Node S3: Local torque 0.401 N·m, local temperature 30.1℃; Node S4: Local torque 0.379 N·m, local temperature 29.5℃; Node S5: Local torque 0.316 N·m, local temperature 28.8℃.

[0051] The signal acquisition and viscosity conversion module performs analog-to-digital conversion on the simulated local torque and temperature signals uploaded from each node, and then first calls the viscosity-torque conversion model for conversion. For node S1, the torque of 0.412 N·m is substituted into the mapping function determined by the Herschel-Balkley constitutive relation and blade geometric parameters to calculate the corresponding original local viscosity as 1.62 Pa·s. Similarly, the original local viscosity for node S2 is 1.48 Pa·s, for node S3 it is 1.56 Pa·s, for node S4 it is 1.45 Pa·s, and for node S5 it is 1.22 Pa·s. Subsequently, temperature drift compensation is performed to uniformly normalize the original local viscosity of each node to a reference temperature of 25℃. Taking node S1 as an example, the current temperature T_m is 31.2℃ (304.35 K), the reference temperature T_ref is 25℃ (298.15 K), and the flow activation energy Ea is 2.85 × 10⁻⁶. 4 Substituting J / mol and R (8.314 J / (mol·K)) into the Arrhenius-type temperature compensation formula, the temperature compensation factor is calculated, yielding a corrected local viscosity of 1.78 Pa·s. Similarly, the compensated local viscosity at node S2 is 1.63 Pa·s, at node S3 it is 1.67 Pa·s, at node S4 it is 1.56 Pa·s, and at node S5 it is 1.33 Pa·s. Thus, the compensated local viscosity signals for the five nodes within this sampling period are 1.78 Pa·s, 1.63 Pa·s, 1.67 Pa·s, 1.56 Pa·s, and 1.33 Pa·s, respectively. These data are transmitted to the weighted data fusion module via an industrial serial bus in the form of data packets with timestamps and node identifiers.

[0052] After receiving the five sets of compensated local viscosity signals, the weighted data fusion module first performs a consistency check. The arithmetic mean of the five sets of data is calculated to be 1.594 Pa·s, and the preset deviation threshold is 20% of the mean, i.e., 0.3188 Pa·s. The deviation of each node is checked one by one: the deviation of node S1 is +0.186 Pa·s, the deviation of node S2 is +0.036 Pa·s, the deviation of node S3 is +0.076 Pa·s, the deviation of node S4 is -0.034 Pa·s, and the deviation of node S5 is -0.264 Pa·s. The absolute value of the deviation of all five nodes does not exceed the threshold of 0.3188 Pa·s, the consistency check is passed, no node is marked as suspicious, and the node abnormality flag signal is set to normal.

[0053] Next, the dynamic calculation of weighted coefficients begins. The weighted data fusion module retrieves historical statistical records for each node. The records show that over the past 30 consecutive sampling periods, the compensated local viscosity variances for nodes S1 to S4 were 0.0009, 0.0011, 0.0008, and 0.0010, respectively, all below the preset variance threshold of 0.0020. This indicates that the slurry flow at these four nodes has become relatively uniform and stable, and the data reliability is high. Node S5, located near the discharge end in the mixing tail section, where the slurry is still in the final stage of axial mixing, has a data variance of 0.0025, slightly higher than the variance threshold. According to the weight adjustment rules, nodes S1 to S4 are assigned higher base weights, while the base weight of node S5 is appropriately reduced. After normalization, the weighted coefficients for the five nodes are determined as w1 = 0.24, w2 = 0.22, w3 = 0.23, w4 = 0.22, and w5 = 0.09. The weighted data fusion module performs a weighted summation calculation, and the current global viscosity is calculated as 1.78 multiplied by 0.24 plus 1.63 multiplied by 0.22 plus 1.67 multiplied by 0.23 plus 1.56 multiplied by 0.22 plus 1.33 multiplied by 0.09, resulting in a value of 1.643 Pa·s. The module packages this global viscosity data along with the weighting coefficients for each node and the node anomaly flag signals, stores it in record position 837 of the circular buffer, and simultaneously pushes the result to the monitoring and early warning module located in the control room via industrial Ethernet.

[0054] After receiving the data for the current period, the monitoring and early warning module plots a continuous curve of global viscosity over time in the trend curve area of ​​its main operation interface. The interface displays the viscosity curve for 837 seconds from the start of stirring to the current moment. The operator observes that between approximately 120 and 150 seconds after stirring starts, the global viscosity gradually decreases from approximately 2.1 Pa·s and stabilizes at around 1.64 Pa·s, indicating that the slurry stirring has entered a uniform steady-state stage. The numerical panel on the right side of the interface displays the current global viscosity as 1.64 Pa·s in prominent font, indicating that the current viscosity is within the preset normal operating range, and the status indicator light is green. Below the operation interface, a bar chart displays the real-time compensated local viscosity of each sensor node and its weighting coefficient in the fusion process. The data bars for nodes S1 to S4 are of similar height and all have high weight markings, while the data bar for node S5 is slightly lower and has a low weight marking. After approximately 40 minutes of continuous operation, the compensated local viscosity output by node S3 in the 2413th sampling cycle suddenly jumped to 2.15 Pa·s, deviating by 32% from the average of 1.63 Pa·s of the other four nodes, exceeding the consistency deviation threshold. In the fusion calculation of the 2413th cycle, the weighted data fusion module excluded node S3, renormalized the weights of the remaining four nodes, and calculated the global viscosity to be 1.62 Pa·s. Simultaneously, it set the node anomaly flag signal corresponding to node S3 to an anomaly and immediately pushed it to the monitoring and early warning module. The monitoring and early warning module switched the icon of node S3 to a flashing red state on the operation interface, popped up an audible and visual alarm window, indicating that node S3 was located 3.1m from the feed end and exhibiting data anomaly, prompting the operator to go to the site for inspection. Upon arrival, operators discovered that hardened slurry lumps adhered to the rotor surface at node S3, causing an excessively high torque reading. After cleaning, the data for node S3 returned to normal starting from the 2460th sampling period. The system automatically reinstated it into the fusion queue and restored the global viscosity to the normal level of 1.64 Pa·s. This embodiment fully demonstrates the system's complete operational mechanism under actual mine backfilling conditions, from multi-point synchronous sensing, rheological model conversion, temperature drift compensation, weighted fusion calculation to anomaly early warning and recovery.

[0055] This invention provides a real-time viscosity detection system for filling slurry in a horizontal mixer. It synchronously collects local signals by axially arranging multiple sets of integrated torque and temperature sensors at the bottom of the mixing tank of the horizontal mixer. The torque signal is converted into the original viscosity using a pre-set viscosity-torque conversion model based on the constitutive equation of non-Newtonian fluids. The system then eliminates the influence of temperature drift based on the temperature-viscosity compensation relationship to generate compensated local viscosity. Finally, a weighted data fusion algorithm dynamically verifies the consistency of data from each node and assigns weights, calculating and outputting a global viscosity signal characterizing the overall state of the slurry.

[0056] Therefore, the real-time viscosity detection system for filling slurry in a horizontal mixer, as described in this invention, solves the problem that existing single-point detection methods cannot obtain a representative real-time overall viscosity of the filling slurry in a horizontal mixer.

[0057] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A real-time viscosity detection system for filling slurry in a horizontal mixer, characterized in that, include: The multi-source sensor array module has at least three sets of viscosity sensor nodes arranged axially at intervals along the bottom of the mixing tank of the horizontal mixer. Each set of nodes integrates a torque detection unit and a temperature sensing unit to synchronously collect local torque simulation signals and local temperature simulation signals of the slurry. The signal acquisition and viscosity conversion module receives the aforementioned signals and performs analog-to-digital conversion. Using a pre-set viscosity-torque conversion model based on the rheological properties of non-Newtonian fluids, it converts the digital torque signal into the original local viscosity signal. Then, based on the temperature-viscosity compensation relationship, it uses the digital temperature signal to compensate for temperature drift, generating a compensated local viscosity signal. The pre-set viscosity-torque conversion model based on the rheological properties of non-Newtonian fluids within the signal acquisition and viscosity conversion module is established in the following way: based on the solid mass concentration, fine particle content ratio, and cementitious material dosage of the filling slurry, it utilizes a pre-calibrated regression function. The yield stress parameter, consistency coefficient parameter, and rheological index parameter are determined. Then, based on the geometric dimensions of the stirring blades and the real-time stirring speed, the stirring speed is mapped to the average shear rate experienced by the slurry. Combining the yield stress parameter, consistency coefficient parameter, and rheological index parameter, the apparent viscosity of the slurry under the current conditions is calculated according to the constitutive relation of non-Newtonian fluids. A correspondence is established between this apparent viscosity and the local torque measured by the torque detection unit, so that the signal acquisition and viscosity conversion module can directly convert the original local viscosity signal from the real-time input local torque digital signal. The formula for calculating the apparent viscosity is as follows: ； in, The apparent viscosity of the filling slurry, For yield stress parameters, Let K be the average shear rate, K be the consistency coefficient parameter, n be the rheological index parameter, and C be the solids mass concentration. This refers to the percentage of fine particles. The real-time stirring speed is given by r, where r is the geometric radius of the stirring blades. This refers to the dosage of cementitious materials; The weighted data fusion module receives the compensated local viscosity signals of all sensor nodes within the same sampling period, performs tests on them, dynamically calculates the weighting coefficients of each node, and generates a global viscosity signal that characterizes the overall state of the slurry in the mixing tank through weighted fusion. The monitoring and early warning module receives global viscosity signals and node anomaly flag signals, displays the global viscosity signals in real time and stores them historically, and outputs a sensor fault early warning signal when the node anomaly flag signal indicates that the sensor node data is abnormal.

2. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 1, characterized in that, Each group of viscosity sensor nodes in the multi-source sensor array module also integrates an embedded microprocessor and a node communication unit. The embedded microprocessor is connected to the torque detection unit and the temperature sensing unit, and is used to drive the torque detection unit and the temperature sensing unit to synchronously perform signal acquisition at a preset sampling frequency, and convert the acquired local torque analog signal and local temperature analog signal into local torque digital signal and local temperature digital signal. The node communication unit is connected to the embedded microprocessor, responds to data requests through an industrial serial communication bus, and encapsulates the local torque digital signal and local temperature digital signal into a data packet before transmitting it to the signal acquisition and viscosity conversion module. Each group of viscosity sensor nodes shares the same clock synchronization pulse to ensure the consistency of sampling time.

3. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 1, characterized in that, The preset temperature-viscosity compensation relationship within the signal acquisition and viscosity conversion module is established based on an Arrhenius-type temperature-viscosity correlation. This relationship describes the quantitative impact of the difference between the local slurry temperature and the preset reference temperature at any sampling time on the slurry viscosity value. The module uses the digital temperature signals synchronously uploaded by each sensor node to determine the current local temperature value. It then multiplies the original local viscosity signal by a temperature compensation factor to generate a compensated local viscosity signal. This temperature compensation factor is jointly determined by the flow activation energy of the filling slurry, the universal gas constant, the preset reference temperature, and the current local temperature value. This ensures that the compensated local viscosity signals output by the system under different slurry temperature conditions are all reduced to the equivalent viscosity value under the preset reference temperature condition. The calculation formula for the temperature compensation factor is as follows: ； in, For temperature compensation factor, R is the activation energy for the filling slurry flow, and R is the universal gas constant. For preset reference temperature, Here is the current local temperature value, n is the rheological index parameter, and K is the consistency coefficient parameter.

4. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 1, characterized in that, The specific method by which the weighted data fusion module performs consistency checks on the compensated local viscosity signals of all sensor nodes within the same sampling period is as follows: calculate the mean of all compensated local viscosity signals, and determine whether the deviation between the compensated local viscosity signal corresponding to each sensor node and the mean exceeds a preset deviation threshold. If the deviation of the compensated local viscosity signal of a certain sensor node exceeds the preset deviation threshold, then the sensor node is marked as a suspicious node in the current sampling period, the compensated local viscosity signal corresponding to the suspicious node is excluded in the current weighted fusion calculation, and the node abnormality flag signal of the suspicious node is set to abnormal and transmitted to the monitoring and early warning module to trigger the output of the sensor fault early warning signal.

5. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 4, characterized in that, The weighted data fusion module dynamically calculates the weighting coefficients for the compensated local viscosity signals of valid sensor nodes that have passed the consistency check. Specifically, it maintains historical data statistics records for each sensor node, including the data variance and fluctuation amplitude of the node over multiple consecutive sampling periods. If the data variance of a sensor node is lower than a preset variance threshold over a preset number of consecutive sampling periods, it indicates that the slurry at the location of the sensor node tends to be uniform, and the weighting coefficient corresponding to the node is increased. If the data fluctuation amplitude of a sensor node exceeds a preset fluctuation threshold, the weighting coefficient corresponding to the node is decreased. After the weighted coefficients are normalized, the weighted data fusion module sums the compensated local viscosity signals of each valid sensor node that has passed the consistency check with their respective normalized weighted coefficients to generate the global viscosity signal for the current sampling period. The calculation formula for the global viscosity signal is as follows: ； in, This is the global viscosity signal. The local viscosity signal after compensation at the i-th node. The normalized weighting coefficient for the i-th node is... Let be the variance of the historical data of the i-th node. This is the weighting coefficient fluctuation correction amount.

6. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 1, characterized in that, The weighted data fusion module is also equipped with a ring-shaped buffer storage area. The ring-shaped buffer storage area is used to store the global viscosity signal generated by each weighted fusion calculation in the order of sampling period, as well as the compensated local viscosity signal, weighting coefficient and node anomaly flag signal corresponding to each sensor node in the sampling period. The storage capacity of the ring-shaped buffer storage area is sufficient to accommodate a preset number of historical data records generated by the system during continuous operation. When the weighted data fusion module receives a data backtracking request from the monitoring and early warning module, it extracts historical data records within a specified time range from the ring-shaped buffer storage area and uploads them to the monitoring and early warning module for viscosity change trend analysis and post-event source tracing of abnormal operating conditions.

7. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 1, characterized in that, After receiving the global viscosity signal and node anomaly flag signal transmitted in real time by the weighted data fusion module, the monitoring and early warning module presents the time evolution of the global viscosity signal on the operation interface in the form of numerical display and trend curve. It also uses a hierarchical color coding method to visually distinguish the global viscosity signal in different numerical ranges. When the node anomaly flag signal indicates that there is a data anomaly in at least one sensor node, the monitoring and early warning module highlights the identifier of the abnormal sensor node and its corresponding local location information on the operation interface. At the same time, it generates a sensor fault early warning signal containing the abnormal node identifier, the time of the anomaly, and the viscosity deviation amplitude, and prompts the operator to perform sensor maintenance and inspection in the form of an audible and visual alarm.

8. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 7, characterized in that, The monitoring and early warning module also continuously judges the abnormal sensor node data indicated by the node abnormality flag signal. If the node abnormality flag signal of the same sensor node is continuously in an abnormal state for a preset number of consecutive sampling periods, the monitoring and early warning module marks the sensor node as a fault node, outputs a sensor fault confirmation signal to the operator through the operation interface, and sends a node disable command to the weighted data fusion module through the communication interface in the subsequent viscosity global calculation process. The command instructs the weighted data fusion module not to introduce the compensated local viscosity signal corresponding to the fault node into the weighted fusion calculation from the next sampling period, and to re-normalize the weighting coefficients of the remaining valid sensor nodes to maintain the continuous output of the global viscosity signal.

9. The real-time viscosity detection system for filling slurry in a horizontal mixer according to claim 1, characterized in that, The monitoring and early warning module also receives externally input slurry quality parameter signals. These signals include at least the slurry solids concentration and the proportion of fine particles. The monitoring and early warning module aligns the slurry quality parameter signals with the global viscosity signal at the corresponding time point according to the timestamp and stores them together. When the global viscosity signal exceeds the preset allowable viscosity range, the monitoring and early warning module automatically compares the current slurry quality parameter signals with the historically stored signals and performs deviation analysis. It then generates control prompts that suggest adjusting the slurry solids concentration and the proportion of fine particles, and displays these prompts to the operator on the interface to assist in the quality control of the filling slurry mixing.

Images