Automatic experiment system for special mortar formula research and development

Abstract

The invention relates to the technical field of general chemical or physical laboratory equipment, and discloses an automatic experiment system for special mortar formula research and development, the system monitors the operation parameters of a stirring unit through a central controller, after the mixing process enters a stable stage, a process fingerprint constructed in real time is compared with a failure fingerprint pattern library, and the failure fingerprint pattern library is compared with a failure fingerprint pattern library; according to the invention, the experimental process is converted from a rigid time sequence to a dynamic process of real-time feedback and resource allocation based on the evolution information of the process, so that the experimental process is more accurate and reliable, and the working efficiency of the experimental process is greatly improved. And the ineffective investment of subsequent time and resources caused by waiting and treating unqualified samples in the initial mixing stage is avoided.

Description

Technical Field

[0001] This invention relates to an automated experimental system for the research and development of special mortar formulations, belonging to the technical field of general chemical or physical laboratory equipment. Background Technology

[0002] Currently, using automated experimental systems to perform high-throughput sample preparation and testing has become a common technical consensus for improving R&D efficiency. These systems can accurately execute a series of preset processes such as material weighing, feeding, and mixing, ensuring the repeatability of experimental operations. Through standardized physical execution, they provide highly consistent samples for subsequent performance evaluation. However, the operation of all such automated systems must follow a more fundamental principle determined by the physicochemical properties of cement-based materials themselves: that is, the series of micro-dynamic evolutions that occur within the material in the first tens of minutes after mixing with water, such as wetting, dispersion, flocculation, and the formation of early hydration products, largely determine the final macroscopic performance. This early dynamic process is manifested through the continuous changes in the rheological state of the slurry.

[0003] However, the working logic of all automated experimental systems is built around a fixed approach of evaluating the endpoint state. The system acts as a faithful executor, strictly following the set program to complete steps such as mixing, molding, and curing. Afterward, the sample to be tested is handed over to external testing instruments as an information black box, passively waiting for an evaluation result on its final performance. The inherent contradiction of this working method is that it knows the decisive role of the early process, but selectively ignores the real-time insight into this process in the system design. As a result, the system will spend hours or even days carefully curing and waiting for samples whose internal structure is destined to fail to form a qualified product within the first hour after mixing. This practice of investing a large amount of machine time, energy consumption, and curing space into an experimental path that is destined to fail is a generally accepted systemic resource misallocation in current automated experimental methods.

[0004] To address this issue, a seemingly straightforward improvement approach is to integrate a dedicated online rheometer or hydration heat analyzer into the automated system to directly measure early process parameters. However, this approach significantly increases the structural complexity and cost of the originally single-function B01L preparation system due to the introduction of a complex G01N analysis module. It also introduces additional challenges related to sample cross-contamination and cleaning / maintenance, thus failing to become a universally applicable solution. Therefore, existing technologies generally suffer from the following limitations: 1. Process information blind spots: In the core mixing stage, the automated system is completely oblivious to the early kinetic evolution processes occurring within the sample that determine its ultimate fate. 2. Rigid workflow: The entire experimental process is a one-way open loop; once started, it cannot be dynamically adjusted or prematurely terminated based on the real-time performance of the sample, lacking necessary process self-reflection and decision-making capabilities. 3. Misaligned resource allocation: A large amount of system resources and time are indiscriminately invested in all samples, failing to prioritize resources for experimental pathways with greater success potential. Therefore, the technical problem to be solved by this invention is how to enhance the inherent ability of an automated system to perceive the early process dynamics of a sample in situ, non-destructively, and at low cost, without adding extra complex analytical hardware or changing the standard hybrid process, by utilizing the most basic execution unit already existing in the automated system, and based on this perception result to make intelligent real-time decisions and optimize the experimental process, thereby transforming the automated experimental system from a passive end-point state evaluator into an active early process identifyer. Summary of the Invention

[0005] This invention provides an automated experimental system for the research and development of special mortar formulations. Its main purpose is to solve the problem that existing automated experimental systems have information blind spots in the mixing stage and cannot effectively identify the early state of the sample, resulting in a lot of time and resources being wasted on samples that are destined to fail.

[0006] To achieve the above objectives, this invention provides an automated experimental system for the research and development of special mortar formulations. The system includes a mixing unit monitoring module and a central controller. The central controller contains a set of operating rules that define the system's operating mode, specifically: Before the mixing process begins, the stirring unit is controlled to run in the reference phase, and the monitoring module is instructed to obtain an operating parameter at this moment as a baseline value. During the mixing process, the operating parameters of the stirring unit are continuously collected to form time series data, and a standard deviation within a rolling time window is calculated based on the time series data. When the calculated standard deviation is first and sustained below a stability threshold used to characterize the end of macroscopic mechanical chaos, a decision process is triggered. In this decision process, time series data collected after this time point are corrected for baseline values ​​and constructed as a process fingerprint. The morphology of the process fingerprint is compared with a failure fingerprint pattern library for similarity. The similarity judgment criteria used for this comparison are adjusted based on the obtained baseline value through a defined mapping relationship. If the similarity of the comparison meets the adjusted similarity judgment criteria, the central controller will execute an instruction to terminate the current mortar sample mixing and subsequent processing process before the end of a planned mixing period.

[0007] Preferably, the operating parameters of the stirring unit are the power consumption values ​​of the stirring motor; the reference stage of operation is no-load operation or pre-stirring operation of dry powder materials.

[0008] Preferably, the central controller performs baseline correction, specifically by subtracting the baseline value from each data point in the time series data.

[0009] Preferably, the central controller is further defined as follows: when the process fingerprint is compared with the failure fingerprint pattern library, the similarity is... Meet the conditions When this occurs, an active analysis process is triggered; among which, and The lower and upper thresholds used to define the fuzzy interval in the similarity judgment conditions are: the active analysis process is as follows: a speed disturbance with a defined waveform and duration is applied to the motor of the stirring unit, and the transient response characteristics caused by the speed disturbance on the operating parameters are monitored simultaneously. The central controller makes the final decision on whether to terminate early based on the analysis results of the transient response characteristics.

[0010] Preferably, the defined mapping relationship is a lookup table or function stored in the central controller, which defines the association between different numerical ranges of the baseline value and a set of different similarity thresholds corresponding to them.

[0011] Preferably, the failure fingerprint pattern library includes multiple failure fingerprint patterns, each of which is a process fingerprint morphology generated by a mortar sample that was ultimately verified as unqualified in historical experiments.

[0012] Preferably, for all mortar samples that have not been prematurely terminated and have completed the entire process, the central controller is further defined to associate the process fingerprint of the mortar sample with its final measured macroscopic performance data and store it in a fingerprint-performance association database.

[0013] Preferably, the central controller is further defined as follows: when it is necessary to reproduce a successful formula on another mixing device, it retrieves the process fingerprint of the successful formula from the fingerprint and performance association database, calculates a total specific energy value by integrating the process fingerprint, and together they form a standard process file; and adopts a closed-loop control method to dynamically adjust the mixing speed of the other mixing device so that the real-time process fingerprint generated on the new device matches the fingerprint in the standard process file, and the termination condition of the mixing process is that the cumulative applied total specific energy value reaches the total specific energy value recorded in the standard process file.

[0014] Preferably, the central controller uses a dynamic time warping algorithm or a cross-correlation algorithm to compare the similarity between the execution process fingerprint pattern and the patterns in the failure fingerprint pattern library.

[0015] Preferably, the central controller is further defined as follows: automatically controlling the system to prepare a standard reference sample and measure its process fingerprint according to a set time period; when the difference between the measured process fingerprint and a stored reference fingerprint exceeds the allowable range, automatically performing global correction on the similarity judgment condition, or outputting a maintenance alarm.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. The central controller monitors and records the time series of operating parameters of the mixing unit during the mixing process in real time, and continuously compares them with the pre-stored failure fingerprint patterns that characterize known non-conforming samples. When the similarity between the real-time acquired process fingerprint and any failure mode reaches a preset condition, the controller will make a decision to terminate the mixing of the current sample and subsequent processing at any time before the end of the mixing process. This transforms the experimental process from a preset rigid sequence to a dynamic process based on the process's own evolution information for real-time feedback and resource allocation, avoiding the subsequent time and resource investment caused by waiting for and processing samples that are destined to be non-conforming in the early stages of mixing.

[0017] 2. This invention also provides a decision triggering mechanism. In the initial stage of the mixing process, this mechanism prioritizes the continuous calculation of the short-term rolling standard deviation of the time series of operating parameters and continuously compares it with a threshold characterizing signal stability. The functional module in the central controller used to compare the process fingerprint with the failure mode library is only activated after the aforementioned rolling standard deviation first falls below the threshold. This design, which uses the statistical stability of the signal as a precondition for starting the core decision function, enables the system to autonomously distinguish between the macroscopic mechanical chaos stage in the initial stage of material mixing and the subsequent chemical kinetic-dominated stage. This ensures that the subsequent morphological-based comparative analysis always occurs within a time window where information is pure and chemical significance is prominent, thus eliminating the possibility of making erroneous termination decisions due to initial physical noise interference.

[0018] 3. When the system identifies that the real-time process fingerprint is similar to both the qualified mode and the failure mode within a preset fuzzy range during the aforementioned decision window, the central controller will switch its control logic and apply a preset, short-term micro-perturbation of the operating parameters to the stirring unit. The system does not rely on external sensors but continues to capture and analyze the transient response characteristics caused by the perturbation on the process fingerprint through the original operating parameter monitoring module. Based on the morphology of the response characteristics, it assists in making the final decision on whether to terminate early. This operating mode, which reuses the execution component (stirring unit) as an active detection tool under specific conditions and interprets two different dimensions of data, passive observation information and active response information, from a single information source (operating parameters), enables the system to actively question and identify the internal microstructure state of the sample through macroscopic appearances when facing situations where the process evolution characteristics are unclear. Attached Figure Description

[0019] Figure 1 This is a flowchart of the dynamic decision-making and intelligent identification process of the system of the present invention; Figure 2 This is a comparison chart of process fingerprint power curves for mortar samples with different properties according to the present invention; Figure 3 This is a schematic diagram of the hardware architecture and information flow of the closed-loop control system of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are only some embodiments of this invention, not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] This invention provides an automated experimental system for the research and development of special mortar formulations. Its overall architecture is built upon a closed-loop control system consisting of a mixing unit monitoring module and a central controller. The mixing unit is a standard laboratory material mixing device. The monitoring module is configured to continuously collect data on one or more operating parameters of the mixing unit. The central controller, as the system's decision-making core, receives and processes the real-time data stream provided by the monitoring module and dynamically adjusts the operation of the mixing unit and the entire experimental process based on process information, according to a set of fixed operating rules. In a specific experimental process, to eliminate the impact of environmental temperature fluctuations on subsequent... To mitigate interference from continuous measurements, the system is configured so that before the mixing process begins, the central controller controls the stirring unit to perform a reference phase of operation, and instructs the monitoring module to acquire an operating parameter as a baseline value during this phase. The reference phase can be either no-load operation or pre-mixing of the dry powder material, both characterized by a constant and low stirring load. For example, the central controller can instruct the stirring unit to run at 60 RPM for 10 seconds without load. During this period, the monitoring module collects the power consumption value of the stirring motor at a frequency of 10 Hz. The central controller then performs an arithmetic average of the collected power values ​​to obtain a baseline value reflecting the current thermodynamic initial state of the system; for instance, at room temperature (20°C). The baseline value measured at that time might be 10.2 W, while at room temperature (30°C)... The baseline value may be 10.8W, which is recorded by the system and used as a benchmark for all subsequent data corrections and decision logic adjustments.

[0022] When materials and water are added to the mixing unit, the mixing process begins. To distinguish between the signal fluctuations caused by the physical movement of materials in the initial stage of mixing and the stable signal dominated by the subsequent chemical hydration reaction, the system employs a decision-triggered gating mechanism. The procedure is as follows: during the mixing process, the central controller continuously collects the operating parameters of the mixing unit to form time-series data, and calculates a standard deviation within a rolling time window based on this data. For example, a rolling time window of 5 seconds is set, and the standard deviation of 50 power data points within this window is calculated in real time. In the initial stage of mixing, the tumbling and adhesion of materials to the walls will cause violent fluctuations in the power signal, with a standard deviation potentially exceeding 5.0W. As the materials gradually form a uniform slurry, the power curve tends to smooth out, and the standard deviation decreases. The system has a preset stability threshold to characterize the end of the physical mixing stage, providing an objective and reproducible basis for decision-triggered activation. The specific value of the stability threshold is determined through the following standardized calibration procedure: First, select no fewer than five representative standard mortar formulations whose final performance has been verified as qualified, and then test them under constant environmental conditions, i.e., a temperature of 23±2°C. At a relative humidity of 50±5%, the mixing process was repeated five times for each formulation, and the total power consumption time series was recorded. Next, for each experiment's power time series data, a 5-second rolling time window was used to calculate its standard deviation, generating a curve showing the standard deviation over time. Subsequently, to objectively identify the turning point of each curve transitioning from a macroscopic mechanical chaos stage to a chemically kinetic-dominated stage, linear fitting was performed on the initial decreasing segment from 0 to 30 seconds and the stable segment 60 seconds before the end of mixing. The projection value of the intersection of the two fitted lines on the vertical axis was defined as the candidate threshold for that experiment. Finally, all candidate thresholds obtained from all experiments were statistically sorted, and the value at the 90th percentile was stored in the central controller as a globally unified stability threshold. When the standard deviation calculated by the central controller is continuously lower than this stability threshold for the first time—for example, below 0.5W for three consecutive seconds—a judgment process is triggered. This mechanism ensures that subsequent decision analysis is conducted within a time window where information stability and chemical significance are prominent.

[0023] After the decision-making process is triggered, the central controller begins constructing a process fingerprint for decision-making. Specifically, the system performs baseline correction on the time-series data collected after the trigger point by subtracting the baseline value obtained at the start of the experiment from each data point in the time-series data. For example, if the currently collected power value is 35.6W and the initial baseline value is 10.2W, the corrected net power value is 25.4W. This operation aims to eliminate systematic deviations caused by ambient temperature and inherent equipment friction, highlighting the resistance changes caused by changes in the rheological properties of the mortar. The corrected time-series data constitutes a process fingerprint characterizing the early hydration kinetics of the current sample. Subsequently, the system enters the real-time decision-making stage, comparing the form of the real-time constructed process fingerprint with a pre-stored failure fingerprint pattern library. This failure fingerprint pattern library includes process fingerprint forms generated by multiple mortar samples that have been verified as unqualified in historical experiments. For example, the false setting pattern is characterized by power exceeding a certain time window after the start of mixing. The peak value is unusually steep, while the segregation and bleeding pattern shows that the power cannot establish effective growth for a long time. The comparison algorithm can use dynamic time warping algorithm or cross-correlation algorithm to measure the similarity of the two time series curves in terms of shape, without being affected by slight scaling or translation on the time axis. It should be noted that the similarity judgment condition used for this comparison is not a fixed value, but is adjusted according to the baseline value obtained in this experiment through a certain mapping relationship. This mapping relationship can be a lookup table or function stored in the central controller. For example, the lookup table can be defined as follows: when the baseline value is in the range of 10.0W to 11.0W, the similarity threshold for triggering the termination decision is 0.95, and when the baseline value is in the range of 11.1W to 12.0W, the threshold is adjusted to 0.90. If the comparison similarity between the current real-time process fingerprint and any pattern in the failure fingerprint pattern library meets the similarity judgment condition after dynamic adjustment, the central controller executes the instruction to terminate the current mortar sample mixing and subsequent processing flow without waiting for the planned mixing time to end.

[0024] If the similarity S between the real-time process fingerprint and the failure mode database falls within a preset fuzzy range, for example, if the condition is met... ,in, and To define the lower and upper thresholds of the fuzzy range and to overcome this decision ambiguity, the central controller is configured to trigger an active analysis process. Specifically, this process involves applying a speed disturbance with a defined waveform and duration to the motor of the stirring unit. For example, a square wave disturbance with an amplitude of ±5 RPM lasting 0.5 seconds is superimposed on a constant speed of 60 RPM. The system, through its existing monitoring module, synchronously monitors the transient response characteristics of the speed disturbance on the operating parameters. For a slurry with a well-developed internal gel network that has initially formed, the response is characterized by a high-amplitude, rapidly decaying peak. Conversely, for a slurry with a loose internal structure, the response is weaker. The central controller then analyzes these transient response characteristics. The analysis results are used to make the final decision on whether to terminate the process early. In the active analysis process, in order to transform the morphological analysis of transient response characteristics into a deterministic decision instruction, its inherent decision partitioning diagram is established and quantified according to the following procedure: Select a stable qualified sample Sample-G, a non-qualified sample Sample-B that will fail, and a critical sample Sample-M whose performance is at the boundary state. Repeated mixing experiments are conducted on these three types of samples. When the mixing reaches the 15th minute, a square wave speed perturbation with an amplitude of ±5 RPM and a duration of 0.5 seconds is uniformly applied. For the transient response generated by each perturbation, the power data within a 2.0-second window after the perturbation occurs is extracted, and the peak amplitude is calculated from it. With response decay time Two quantitative metrics, one of which, decay time, is defined as the time from the peak point until the signal amplitude first decays to 37% of the peak value. The time elapsed after doubling; finally, collect all the experimental data. The numerical pairs and their corresponding final sample performance labels are used as a training dataset and input into a linear support vector machine (SVM) classifier for training. After the classifier is trained, it generates a dataset of the form... The decision function, and its decision rule, namely... Continue the process The execution of the termination command at the appropriate time constitutes the specific implementation of the decision partitioning diagram within the central controller.

[0025] For all mortar samples that were not prematurely terminated and completed the entire process, the central controller will associate the process fingerprint of the mortar sample with its final measured macroscopic performance data and store it in a fingerprint-performance association database, allowing the system's knowledge base to be continuously improved during use. Furthermore, when a successful formula needs to be reproduced on another mixing device, the central controller retrieves the process fingerprint of that successful formula from the fingerprint-performance association database, integrates the process fingerprint over time, divides it by the sample mass, and calculates a total specific energy value. This process fingerprint and the total specific energy value together constitute a standard process file. During reproduction, the system employs a closed-loop control method. The system dynamically adjusts the stirring speed of another stirring device to ensure that the real-time process fingerprint generated on the new device matches the fingerprint in the standard process file. The termination condition for the stirring process is that the cumulative applied specific energy value reaches the total specific energy value recorded in the standard process file. To ensure the long-term stability of the system, the central controller is also configured to automatically control the system to prepare a standard reference sample and measure its process fingerprint according to a set time period, such as once a week. When the difference between the measured process fingerprint and a stored reference fingerprint exceeds the allowable range, it indicates that the mechanical performance of the stirring unit may have drifted. At this time, the system can automatically perform global correction of the similarity judgment condition or output a maintenance alarm.

[0026] Example 1: In an experimental project aimed at high-throughput screening to develop a rapid-setting, high-early-strength special mortar, an automated experimental system was deployed to continuously prepare and evaluate samples of a series of formulations. A technical challenge in this scenario is that some formulations may experience false setting or segregation in the early stages of the mixing process due to interactions between their components. If a fixed-duration experimental procedure is followed, the system will allocate machine time and subsequent curing resources to these samples that have already failed in the early stages. When the system begins preparing a new formulation sample numbered Sample-073, its operation is as follows: The system first performs a reference stage, i.e., pre-mixing the dry powder material added to the mixing unit for 30 seconds. The central controller obtains the average power consumption value of this stage through the monitoring module and... The recorded value serves as the baseline for this experiment. This step provides a dynamic reference coupled with the current equipment status and ambient temperature for subsequent judgment processes. Subsequently, the system adds a fixed amount of water according to the formula and begins mixing. In the first two minutes after mixing begins, the power consumption value collected by the monitoring module exhibits significant random fluctuations. The decision-triggered gating mechanism in the central controller is activated at this stage, continuously calculating the standard deviation of the power signal within a 5-second rolling time window. Only when the standard deviation is lower than the preset stability threshold for the first time for 3 consecutive seconds does the mechanism send a trigger signal to the judgment process module. This mechanism isolates the subsequent morphological comparison from the physical noise in the early stage of mixing, making the input for morphological comparison a time series data that has been filtered out of physical mixing interference and mainly reflects the chemical kinetic evolution.

[0027] After the decision process is triggered, the central controller begins to construct a process fingerprint from the real-time collected and baseline-corrected time-series data and compares it with a failure fingerprint pattern library. At the 11-minute mark of the mixing process, the similarity between Sample-073's process fingerprint and a failure fingerprint pattern marked as false coagulation in the library reaches a high value. At this point, instead of comparing this similarity to a fixed threshold, the central controller invokes a similarity judgment condition adjusted based on the initial baseline value. Because the laboratory ambient temperature was slightly higher at the start of the experiment, resulting in a correspondingly higher initial baseline value, this judgment condition, based on a preset lookup table, lowers the similarity threshold for triggering termination from the usual 0.90 to 0.85. Given the currently calculated similarity... The similarity had exceeded the adjusted threshold of 0.85, and the central controller made a termination decision. This working method, which combines signal stability quality inspection with environmental adaptive threshold adjustment, provides a mechanism to synchronously correct the judgment benchmark when using changing signals for judgment. This transforms the system's working mode from passive end-point evaluation to an active process with early process identification capabilities. The central controller then executed the instruction to terminate the current mortar sample mixing and subsequent processing flow. The system automatically discharged Sample-073 as waste and immediately began the preparation process of the next sample, Sample-074. The time and space resources previously occupied by Sample-073 for subsequent mixing, molding, and curing were immediately released.

[0028] Example 2: To objectively verify the effectiveness of the automated experimental system of this invention in early identification and process termination of mortar samples with different performance characteristics during the R&D workflow, the following comparative experiment was designed and executed; the experimental platform adopted the automated experimental system described in the aforementioned specific embodiment, its mixing unit was a planetary mortar mixer with a speed control accuracy of ±1RPM, its monitoring module was a power sensor connected to the mixing motor, the data acquisition frequency was set to 10Hz, and the power resolution was 0.1W; a failure fingerprint pattern library containing two failure modes, false setting and segregation bleeding, was pre-loaded into the central controller; 10 sets of pre-verified, high-performance materials were selected. Various special mortar formulations, numbered P-01 to P-10, are used. Among them, P-01 to P-04 are formulations that meet the final performance requirements, P-05 to P-07 are formulations that will experience false setting during mixing, and P-08 to P-10 are formulations that will experience segregation and bleeding. The experiment is divided into a control group and an experimental group. The control group adopts a fixed-duration working mode, that is, the early identification and termination functions of the system are disabled. All 10 formulation samples are subjected to a complete mixing process with a duration of 30.0 minutes. The experimental group uses the automated experimental system of this invention. The stability threshold in its decision trigger gating mechanism is set to 0.5W, and the other parameters are consistent with those in the specific implementation method.

[0029] The experimental results showed that in the control group, all 10 sample groups underwent a complete 30.0-minute mixing process. In the experimental group, for samples P-01 to P-04, whose expected performance was satisfactory, the system did not trigger an early termination command and completed the entire 30.0-minute mixing process, with the judgment results consistent with the expected performance. In contrast, for samples P-05, P-06, and P-07, which were expected to experience false coagulation, the system made an early termination decision at 12.4 minutes, 14.8 minutes, and 13.1 minutes of mixing, respectively. Similarly, for samples P-08, P-09, and P-10, which were expected to experience segregation and bleeding, the mixing process ended at 16.7 minutes, 18.5 minutes, and 18.5 minutes, respectively. The experiment was terminated by the system at 19.2 minutes. The system's judgment results for the experimental group were consistent with the expected performance of all 10 groups of samples. The data from this experiment show that the automated experimental system can effectively identify the performance evolution trend of samples in the early stage of material mixing and automatically terminate the subsequent preparation process of those samples that are judged to be unqualified. The underlying mechanism is that the process fingerprint morphology generated by the unqualified samples in the early stage of mixing reaches the similarity judgment condition of the corresponding pattern in the failure fingerprint pattern library after dynamic adjustment, thereby triggering the termination command. Compared with the fixed duration working mode, the system reduces the invalid machine time allocated to unqualified samples by 35% to 58%, improving the time utilization rate of R&D equipment.

[0030] Example 3: This example combines Figures 1 to 3 This describes an automated experimental system for the research and development of special mortar formulations, such as... Figure 1 As shown, the process begins with the start of the experiment / introduction of dry powder materials, followed by a reference phase to obtain baseline values ​​to correct for environmental influences. After the formal mixing starts, the system uses a decision-triggered gating mechanism to compare the calculated rolling standard deviation with a preset stability threshold to determine whether the mixing process has escaped macroscopic mechanical chaos. When the standard deviation is less than the stability threshold, the subsequent judgment process is triggered, namely, entering the process fingerprint construction stage. The data corrected by the baseline value is used to construct a process fingerprint characterizing the early hydration kinetics of the sample. This process fingerprint is compared and judged with the typical fingerprint morphology of unqualified samples in the failure fingerprint pattern library. If the similarity meets the termination condition dynamically adjusted by the baseline value, it is judged as an unqualified sample and the early termination process is executed to release machine time and resources. If the similarity does not meet the termination condition, the sample enters the entire mixing process and becomes a qualified sample. Its process fingerprint and final performance data are stored in the fingerprint and performance association database for knowledge accumulation and process reproduction. When the similarity is in the fuzzy range, the active analysis process is triggered. By applying micro-perturbations of rotation speed and analyzing transient response characteristics, it provides auxiliary basis for the final decision.

[0031] like Figure 2 As shown in the figure, the horizontal axis represents time (seconds), and the vertical axis represents the corrected power (W). The figure displays three typical process fingerprint curves. The solid line, marked as a qualified sample, shows a steady and continuous increase in power value after the start of mixing, eventually stabilizing at a relatively high plateau, reflecting the dynamic process of stable formation of the internal structure of the slurry. The thick dashed line, marked as a false-coagulated sample, shows an abnormally steep increase in power value after about 300 seconds of mixing, significantly higher than the normal level of the qualified sample, indicating a sharp loss of slurry fluidity. The dotted line, marked as a segregated and bleeding sample, shows a low power value throughout the mixing process with weak growth, reflecting that the material failed to form a uniform and stable gel structure.

[0032] like Figure 3 As shown, the system architecture is centered on a central controller, which includes a decision-making module, data processing unit, storage unit, and operating rule fixing module. The central controller sends parameter acquisition commands to the monitoring module, which monitors the power of the mixing unit through its internal power sensor. The mixing unit includes a feeding port, mixing container, mixing motor, and control panel. The monitoring module sends the collected data back to the central controller in real-time at a frequency of 10Hz. The central controller analyzes and makes decisions based on the received data stream and sends a pattern comparison request to the database system. This database system contains a failure fingerprint pattern library and a fingerprint performance association library, and returns the similarity comparison results to the central controller. Finally, based on the operating rules and the comparison results, the central controller sends control commands such as speed control to the mixing motor in the mixing unit, thus forming a closed-loop dynamic decision and control system.

[0033] Example 4: When an automated experimental system was first deployed in a new special mortar R&D project, due to a lack of historical data for this material system, the fuzzy range upper and lower limit thresholds used by the central controller to trigger the active analysis process... and Furthermore, the key parameters of the proactive analysis process are all in an undefined state. To solve this problem, the system needs to perform a standardized initial parameter calibration and decision logic quantification procedure before it is put into regular use.

[0034] The first step of this procedure is to calibrate the boundaries of the fuzzy interval. The operator needs to prepare three types of standard samples in advance: a stable, qualified sample (Sample-G), a non-qualified sample (Sample-B) that will fail, and a critical sample (Sample-M) whose performance is at the boundary. With the early termination function disabled, the system performs five repeated mixing experiments on each of these three types of samples, recording the complete process fingerprint for each sample. The central controller then calculates and statistically analyzes the similarity distribution data, including the similarity between qualified and non-qualified samples, and the similarity between the critical sample and the qualified sample, and between the critical sample and the non-qualified sample. Based on this data distribution, the system can determine the initial threshold, and accordingly... The lower boundary value of the similarity distribution between critical and qualified samples is set to 0.80, and... The upper boundary value of the similarity distribution between critical and non-compliant samples is set to 0.90.

[0035] The second step of this procedure is to determine the rotational speed perturbation parameters applied in the active analysis process. The determination method aims to ensure that the signal-to-noise ratio of the response signal meets the requirements while avoiding interference with the hydration process of the sample itself. The calibration method is to use the aforementioned critical sample Sample-M for testing. At the 15th minute of the mixing process, the central controller applies square wave rotational speed perturbations of different amplitudes and durations according to the preset gradient parameter table, and records the transient response caused by each perturbation in terms of power consumption parameters. The final selected parameter combination is an amplitude of ±5 RPM and a duration of 0.5 seconds. The criteria for judgment are that the signal-to-noise ratio of the peak transient response caused by these parameters is greater than 10 dB, and after the perturbation ends, the overall process fingerprint of the sample can recover to the trend line before the perturbation within 20 seconds.

[0036] The third step of this procedure is to quantify the analysis logic of transient response characteristics. When the active analysis process is triggered and the aforementioned calibrated speed disturbance is applied, the central controller executes the following algorithm steps: extracting power consumption data within a 2.0-second time window after the disturbance occurs; subtracting the average power value of the 1.0 second period before the disturbance from this data segment to obtain the net response signal; and calculating two quantized indicators from this net response signal, including peak amplitude. With response decay time The decay time is defined as the time elapsed from the peak point until the signal amplitude first decays to 37% of the peak value; finally, the calculated ( , The numerical value is compared with a decision zoning map established by calibrating the sample. If the point falls in the qualified zone, the system continues to execute the mixing process; if it falls in the unqualified zone, the system executes an early termination command. By executing the above procedure, the automated experimental system completes the parameter calibration under the new material system, providing a quantitative decision basis for its subsequent routine operation.

[0037] Example 5: To ensure the accuracy of the automated experimental system's decision-making under varying laboratory ambient temperatures, the mapping relationships stored internally for adjusting similarity judgment conditions need to be calibrated offline. This procedure is executed in a temperature-controlled chamber. First, the automated experimental system is placed in the chamber, and an initial temperature point is set. After the system temperature reaches equilibrium with the ambient temperature, a reference phase is run to record the baseline value corresponding to this temperature point. Subsequently, a mixed experiment is conducted on the qualified sample Sample-G, the unqualified sample Sample-B, and the critical sample Sample-M, respectively, to determine the similarity judgment conditions that can correctly distinguish the critical sample at this baseline value. The above steps are repeated, that is, at multiple different temperature points, the corresponding baseline values ​​and similarity judgment condition value pairs are sequentially obtained, and these data pairs are filled into a lookup table in the central controller.

[0038] To enable the system's failure fingerprint pattern library to be expanded and optimized during use, the system is also equipped with a set of pattern update and maintenance procedures. In the routine experimental process, for each sample that is ultimately identified as unqualified after completing the entire process, the central controller will archive its process fingerprint and final performance label together. When the number of newly added unqualified samples with the same performance label reaches a preset number, the central controller will automatically perform a cluster analysis on the process fingerprints of these samples. If the analysis results show that there is a fingerprint morphology cluster with high cohesion, the system will calculate a central fingerprint for the newly discovered cluster and add the central fingerprint as a new entry to the failure fingerprint pattern library.

[0039] Example 6: To apply an established standard process profile to large-scale experiments or reproducible production using a new batch of raw materials, a standardized pre-deployment verification and adaptive adjustment procedure needs to be executed to ensure that the consistency of the mixing process is not affected by differences in the physical properties between batches of raw materials. The procedure is as follows: the standard process profile to be verified is specified in the system, and the identification information of the new batch of raw materials is loaded; the system then enters the verification mode and uses a closed-loop control method to mix the samples using the new batch of raw materials. The control objective is to match the real-time generated process fingerprint morphology and total specific energy value with the target values ​​recorded in the standard process profile; during the mixing process, the central controller synchronously calculates the cumulative average between the currently generated process fingerprint and the target process fingerprint. The system calculates the root mean square error (RMSE) and compares it with a preset process fidelity threshold at the end of the mixing process. If the calculated RMSE is lower than the process fidelity threshold, the current standard process file is deemed applicable to the new batch of raw materials. If the RMSE exceeds the threshold, it indicates that the new batch of raw materials has an impact on the mixing process. In this case, the system automatically saves the actual process fingerprint generated in this verification mode, as well as the final total specific energy value, as a new standard process file associated with the identification information of the new batch of raw materials. Through this verification and adaptive adjustment procedure, the system generates corresponding standardized mixing process guidelines for different batches of raw materials, providing traceable consistency assurance for subsequent routine experiments or production tasks.

[0040] After the automated experimental system has been running continuously for a set time period, in order to verify and maintain the long-term stability of its measurement benchmark, the central controller automatically triggers a system self-test and calibration procedure during the intervals between experimental tasks. In this procedure, the system first calls a standard reference sample formula stored internally and automatically controls the mixing and feeding units to complete the preparation of the standard reference sample. Subsequently, the system mixes the sample according to the standard procedure and collects its complete real-time process fingerprint. The central controller uses a dynamic time warping algorithm to compare the newly measured process fingerprint with the benchmark fingerprint stored during the initial calibration of the system for morphological similarity and calculates a drift value. If the drift value is lower than the upper limit of the preset allowable range, the system self-test passes. If the drift value exceeds the allowable range, the central controller outputs a maintenance alarm according to preset rules, or calculates a set of correction coefficients based on the magnitude and direction of the drift value and makes global compensation adjustments to the similarity judgment conditions.

[0041] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An automated experimental system for the research and development of special mortar formulations, the system comprising a mixing unit monitoring module and a central controller, characterized in that, The central controller has a set of operating rules that define how the system operates, including: Before the mixing process begins, the stirring unit is controlled to run in the reference phase, and the monitoring module is instructed to obtain an operating parameter at this moment as a baseline value. During the mixing process, the operating parameters of the stirring unit are continuously collected to form time series data, and a standard deviation within a rolling time window is calculated based on the time series data. When the calculated standard deviation is first and sustained below a stability threshold used to characterize the end of macroscopic mechanical chaos, a decision process is triggered. In this decision process, time series data collected after this time point are corrected for baseline values ​​and constructed as a process fingerprint. The morphology of the process fingerprint is compared with a failure fingerprint pattern library for similarity. The similarity judgment criteria used for this comparison are adjusted based on the obtained baseline value through a defined mapping relationship. If the similarity of the comparison meets the adjusted similarity judgment criteria, the central controller will execute an instruction to terminate the current mortar sample mixing and subsequent processing process before the end of a planned mixing period.

2. The automated experimental system for developing special mortar formulations according to claim 1, characterized in that, The operating parameters of the mixing unit are the power consumption values ​​of the mixing motor; the reference stage of operation is no-load operation or pre-mixing operation of dry powder materials.

3. The automated experimental system for developing special mortar formulations according to claim 1, characterized in that, The central controller performs baseline correction, specifically by subtracting the baseline value from each data point in the time series data.

4. The automated experimental system for developing special mortar formulations according to claim 1, characterized in that, The central controller is further defined as: when the process fingerprint is compared with the failure fingerprint pattern library, the similarity is... Meet the conditions When this occurs, an active analysis process is triggered; among which, and The lower and upper thresholds used to define the fuzzy interval in the similarity judgment conditions are: the active analysis process is as follows: a speed disturbance with a defined waveform and duration is applied to the motor of the stirring unit, and the transient response characteristics caused by the speed disturbance on the operating parameters are monitored simultaneously. The central controller makes the final decision on whether to terminate early based on the analysis results of the transient response characteristics.

5. The automated experimental system for developing special mortar formulations according to claim 1, characterized in that, The defined mapping relationship is a lookup table or function stored in the central controller, which defines the association between different numerical ranges of the baseline value and a set of different similarity thresholds.

6. The automated experimental system for developing special mortar formulations according to claim 1, characterized in that, The failure fingerprint pattern library includes multiple failure fingerprint patterns. Each failure fingerprint pattern is a process fingerprint morphology generated by a mortar sample that was ultimately verified as unqualified in historical experiments.

7. The automated experimental system for developing special mortar formulations according to claim 1, characterized in that, For all mortar samples that were not terminated prematurely and completed the entire process, the central controller is also defined to associate the process fingerprint of the mortar sample with its final measured macroscopic performance data and store it in a fingerprint-performance association database.

8. The automated experimental system for developing special mortar formulations according to claim 7, characterized in that, The central controller is further defined as follows: when a successful formula needs to be reproduced on another mixing device, it retrieves the process fingerprint of the successful formula from the fingerprint and performance association database, calculates a total specific energy value by integrating the process fingerprint, and together they form a standard process file; and adopts a closed-loop control method to dynamically adjust the mixing speed of the other mixing device so that the real-time process fingerprint generated on the new device matches the fingerprint in the standard process file, and the termination condition of the mixing process is that the cumulative applied total specific energy value reaches the total specific energy value recorded in the standard process file.

9. The automated experimental system for developing special mortar formulations according to claim 1, characterized in that, The central controller uses a dynamic time warping algorithm or a cross-correlation algorithm to compare the similarity between the execution process fingerprint pattern and the patterns in the failure fingerprint pattern library.

10. An automated experimental system for the research and development of special mortar formulations according to claim 1, characterized in that, The central controller is also defined as follows: according to a set time period, the automatic control system prepares a standard reference sample and measures its process fingerprint; when the difference between the measured process fingerprint and a stored baseline fingerprint exceeds the allowable range, the similarity judgment condition is globally corrected, or a maintenance alarm is output.

Images