Industrial robot task cooperative scheduling method for intelligent factory

Abstract

The invention discloses an industrial robot task cooperative scheduling method for an intelligent factory, and relates to the technical field of industry, and the method comprises the steps: firstly, in a bromine extraction process, obtaining key core parameters, and based on the key core parameters, judging whether each node has parameter mutation in a conveying and process main process so as to identify a window period encryption signal; the window period encryption signal is used for determining a medium load integral new period, an automatic sampling new period and a robot regulation and control scheduling new period; secondly, on the basis of complete data in a historical time period, a historical load characteristic curve is constructed, the historical load characteristic curve is used for identifying a real-time load change stage corresponding to a current dynamic load state, and the real-time load change stage comprises a rapid growth stage and a slow saturation stage; and finally, according to the real-time load change stage and in combination with a historical load characteristic curve, identifying the offset degree of the current dynamic load state so as to trigger a dual cooperative adjustment mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industry, in particular to an industrial robot task cooperative scheduling method for intelligent factory. BACKGROUND

[0002] In the intelligent factory of the bromine extraction process, the industrial robot task cooperative scheduling needs to be carried out around the whole process of bromine production, covering the core links such as raw material conveying oxidation (such as seawater, chlorine oxidation, hydrochloric acid oxidation), multi-point sampling detection (such as blowout tower air volume, absorption tower outlet sodium bromate and sodium bromide ratio, distillation tower temperature, bromine purity, etc.), process control operation (such as adding raw materials, replacing raw materials) and the like. Through task allocation, timing coordination and dynamic adjustment among robots, efficient operation of the bromine extraction process is realized.

[0003] At present, the task scheduling of industrial robots in the bromine extraction process mostly adopts a fixed cycle mode, that is, sampling, flow regulation and other operations are performed at a preset frequency. This mode has significant limitations: on the one hand, the bromine concentration of seawater is easily affected by factors such as season and climate and suddenly fluctuates. Fixed cycle sampling and detection are difficult to capture parameter mutations in time, resulting in a lag in the adjustment of the robot's raw material conveying amount, which in turn affects the efficiency of the bromine extraction process. On the other hand, the bromine extraction process has different stages, and the fixed task scheduling logic is difficult to adapt to the blowout characteristics of different stages, which easily causes medium overload leakage or resource waste caused by premature replacement.

[0004] When the bromine concentration suddenly decreases, the fixed sampling cycle is difficult to trigger flow regulation in time, resulting in insufficient input and a sharp drop in blowout efficiency, which leads to an extension of the production cycle. When the concentration suddenly rises, the lagging adjustment may cause the blowout to be saturated in advance, and bromide ions are leaked with wastewater, which not only reduces the recovery rate but also pollutes the environment. In addition, misjudgment in the bromine extraction process stage will cause robot task mismatch, such as still scheduling sampling at the frequency of the rapid growth stage in the slow saturation stage, which not only increases the invalid work of the robot, but also may cause decision-making errors due to data redundancy, ultimately causing problems such as increased energy consumption and fluctuating product purity. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides an industrial robot task cooperative scheduling method for intelligent factory, which solves the problems in the above background art.

[0006] To achieve the above purpose, the present application is implemented by the following technical scheme: an industrial robot task cooperative scheduling method for intelligent factory, comprising the following steps, In the bromine extraction process, key core parameters (including temperature, seawater bromine content, chlorine dosage, sodium bromate and sodium bromide ratio, bromine purity, etc.) are obtained. Based on the key core parameters, it is determined whether parameter mutation occurs at each node along the conveying and process main flow to identify the window period encryption signal. The window period encryption signal is used to determine the medium load integration new period, the automatic sampling new period, and the robot control and scheduling new period. Based on the complete data in the historical period, a historical load characteristic curve is constructed. The historical load characteristic curve is used to identify the real-time load change stage corresponding to the current dynamic load state. The real-time load change stage includes a rapid growth segment and a slow saturation segment. According to the real-time load change stage, and in combination with the historical load characteristic curve, the offset degree of the current dynamic load state is identified to trigger a dual cooperative adjustment mechanism. The dual cooperative adjustment mechanism is used to adjust the frequency converter through the industrial robot and identify the start of the medium replacement operation.

[0007] Preferably, in the bromine extraction process, multiple node automatic sampling valves and bypass sampling pipes are arranged along the conveying and process main flow. The detection sample is sent into the colorimetric analyzer through the conveying pump to periodically detect the key core parameters of different process segments. The volumetric flow rate at the inlet of the bromine extraction process tower is obtained through a thermometer and an electromagnetic flowmeter detection device. The key core parameters (including temperature, seawater bromine content, chlorine dosage, sodium bromate and sodium bromide ratio, bromine purity, etc.) are obtained by summarizing the key core parameters of different process segments in time sequence. The relevant parameter data of nodes such as blowout tower, absorption tower, and distillation tower are extracted from the relevant parameter data and marked as pretreatment information. According to the pretreatment information, the change of each node at the adjacent monitoring time point is analyzed to obtain the change rate of the corresponding node. Based on the change rate of the corresponding node, it is determined whether the corresponding node has a parameter mutation to identify the window period encryption signal.

[0008] Preferably, based on the change rate of the corresponding node, it is determined whether the corresponding node has a parameter mutation, including: The change rate sequence of the corresponding node in the historical data is extracted to form a historical sequence. Based on the historical sequence, the mean value of the change rate and the standard deviation of the change rate are determined. If , it is determined that there is a parameter mutation at the corresponding node. At this time, the scheduling instruction is triggered. Otherwise, it is determined that there is no parameter mutation at the corresponding node. At this time, the scheduling instruction is not triggered. Wherein, is the change rate of the i-th node at time t, is the mean value of the change rate, is the standard deviation of the change rate, and k is a constant. determining the time stamp of the parameter mutation and marking as a phase point, based on the phase point, obtaining the dynamic load of the blow tower, the absorption tower and the distillation tower; Preferably, the window period encryption signal comprises: According to the phase point, the scheduling instruction is sent to the centralized scheduling optimization controller, and the centralized scheduling optimization controller obtains a self-adaptive scaling factor according to the abnormal degree of the parameter mutation of the corresponding node, the self-adaptive scaling factor is used to quantify the scale of dynamic regulation, and the centralized scheduling optimization controller is used to dynamically control the medium load integration period, control the automatic sampling period and control the robot regulation scheduling period; The medium load integration period, the automatic sampling period and the robot regulation scheduling period are uniformly compressed by the self-adaptive scaling factor to obtain a new period at the phase point, and the new period comprises a medium load integration new period, an automatic sampling new period and a robot regulation scheduling new period; The new period is taken as the window period encryption signal, and the window period encryption state is maintained until it is monitored that there is no parameter mutation in the corresponding node, the window period encryption state is stopped, and the new period is restored to the original period.

[0009] Preferably, based on the complete data in the historical period, a historical load characteristic curve is constructed, and the historical load characteristic curve is used to identify the real-time load change phase corresponding to the current dynamic load state after the step of identifying the real-time load change phase. Collecting complete data of the medium dynamic load changing with time under the same process condition as the current existing process condition in the historical bromine extraction process; According to the characteristics of the bromine extraction process, combining the complete data, constructing the historical load characteristic curve through a linear function and an exponential function, and dividing the bromine extraction process into a rapid growth segment and a slow saturation segment; In the current dynamic medium load integration accumulation process, the dynamic load of the blow tower, the absorption tower and the distillation tower at the current time is determined; Taking the current time as a starting point, data of the same time span as the historical load characteristic curve is intercepted forwardly; The growth slope of the current dynamic load state is calculated; The growth slope of the current dynamic load state is compared with the slope range of each stage in the historical load characteristic curve to determine the real-time load change phase.

[0010] Preferably, determining the real-time load change phase comprises: If the growth slope of the current dynamic load state is in the slope range of the rapid growth segment, it is determined that the current bromine extraction process is in the rapid growth segment; If the growth slope of the current dynamic load state is in the slope range of the slow saturation segment, it is determined that the current bromine extraction process is in the slow saturation segment; The fast growth section and the slow saturation section are combined to form a real-time load change stage.

[0011] Preferably, according to the real-time load change stage and in combination with the historical load characteristic curve, the offset degree of the current dynamic load state is identified to trigger a dual collaborative adjustment mechanism, including: According to the real-time load change stage, the historical dynamic load corresponding to the time stamp of the current dynamic load state on the historical load characteristic curve is determined and recorded as a predicted dynamic load; By comparing the predicted dynamic load with the current dynamic load state, a deviation value is determined. If the deviation value exceeds a preset deviation threshold, the dual collaborative adjustment mechanism is triggered.

[0012] Preferably, the dual collaborative adjustment mechanism includes: Multi-dimensional cycle adjustment; if the time corresponding to the current dynamic load state is in a window cycle encryption state, the window cycle encryption state is maintained; if the time corresponding to the current dynamic load state is not in the window cycle encryption state, a scheduling instruction is triggered again to realize the window cycle encryption state. Notification task scheduling adjustment: according to the deviation value, the task of the industrial robot is adjusted in advance, specifically: when the real-time load change stage is the fast growth section, the industrial robot is dominated to perform a frequency converter adjustment operation; when the real-time load change stage is the slow saturation section, the industrial robot is dominated to start a medium replacement operation in advance or delay; The multi-dimensional cycle adjustment and the notification task scheduling adjustment are combined to obtain the dual collaborative adjustment mechanism.

[0013] Preferably, when the real-time load change stage is the slow saturation section, the industrial robot is dominated to start a medium replacement operation in advance or delay, including: If the current dynamic load state is higher than the predicted dynamic load, a pre-start instruction is sent to a medium replacement robot cluster to prepare for the advance tower replacement operation; the medium replacement robot cluster includes a plurality of groups of industrial robots. If the current dynamic load state is not higher than the predicted dynamic load, the current dynamic load state is integrated in time according to the multi-dimensional cycle adjustment, and a time stamp of the delayed start of the medium replacement operation is determined according to the integral calculation result and a maximum load capacity; the maximum load capacity refers to the maximum load capacity that the medium can bear.

[0014] The present application provides an industrial robot task collaborative scheduling method for a smart factory, which has the following beneficial effects: (1) By identifying the parameter mutation trigger window period encryption signal, the medium load integral, automatic sampling and robot control scheduling period dynamic compression, ensure the rapid capture of bromine concentration sudden rise / sudden drop, solve the problem of traditional fixed cycle scheduling response lag to process fluctuation. Based on the historical load characteristic curve, the rapid growth section and the slow saturation section are divided, so that the robot task scheduling and the bromine extraction process stage are accurately matched, and the medium overload or resource waste caused by one-size-fits-all scheduling is avoided. Through the dual synergy mechanism of window period encryption and robot task adaptation, when the dynamic load deviates from the expected, the raw material input can be corrected in real time through the frequency converter adjustment, and the medium replacement operation can be started in advance, so that the bromine recovery rate can remain stable, and the energy consumption is further reduced. The traditional scheduling takes sample and adjusts according to fixed period, and when the parameter mutates, the relatively optimal adjustment opportunity is easily missed, resulting in decreased bromine extraction process efficiency or bromide ion leakage. The change rate of node concentration is calculated by self-adaptive statistical method, and when the change rate exceeds the threshold value, the window period encryption signal is triggered to compress the related period. The slope range of each stage is determined by the historical load characteristic curve, the current dynamic load state growth slope is compared with the historical range, and the real-time stage is accurately identified. For example, the robot increases the raw material flow in the rapid growth section to fully utilize the active sites; the medium replacement robot cluster is started in advance in the slow saturation section to avoid overload leakage. Schedule the robot: adjust the frequency converter to stabilize the bromide ion input in the rapid section, and advance / defer the tower replacement to balance efficiency and cost in the slow section. In summary, the present application overcomes the limitations of the traditional fixed scheduling mode through the synergistic design of dynamic sensing, stage adaptation and dual adjustment, further improves the stability, resource utilization and environmental protection of the bromine extraction process, and provides an efficient solution for robot collaborative scheduling in intelligent factories for complex chemical scenes.

[0015] (2) The beneficial effects of the present application are: first, breaking the hysteresis of fixed period scheduling, avoiding the delay of adjustment when the concentration suddenly rises / suddenly drops through the window period encryption triggered by parameter mutation; second, eliminating the waste of resources caused by stage misjudgment, the flow optimization of the rapid growth section improves the utilization rate of medium active sites, and the automatic control of the slow saturation section avoids the waste of industrial extraction caused by not replacing in time; third, when the deviation exceeds the limit, the data collection density can be strengthened, and the robot can be scheduled accordingly, so that the process fluctuation amplitude can be controlled. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A step diagram of a kind of industrial robot task collaborative scheduling method for intelligent factory of the present application. DETAILED DESCRIPTION

[0017] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the protection scope of the present application.

[0018] Embodiment 1 Please refer to Figure 1 The present application provides an industrial robot task coordination scheduling method for a smart factory, comprising the following steps, In the bromine extraction process, key core parameters (including temperature, seawater bromine content, chlorine dosage, sodium bromate and sodium bromide ratio, bromine purity, etc.) are obtained, based on the key core parameters, it is judged whether parameter mutation occurs at each node along the conveying and process main flow, to identify the window period encryption signal, the window period encryption signal is used to determine the medium load integration new period, the automatic sampling new period and the robot control scheduling new period; Based on the complete data in the historical period, a historical load characteristic curve is constructed, the historical load characteristic curve is used to identify the real-time load change stage corresponding to the current dynamic load state, the real-time load change stage includes a rapid growth segment and a slow saturation segment; According to the real-time load change stage, and combining the historical load characteristic curve, the offset degree of the current dynamic load state is identified, to trigger a double coordination adjustment mechanism, the double coordination adjustment mechanism is used to adjust the frequency converter through the industrial robot and identify the start of the medium replacement operation.

[0019] In this embodiment, by monitoring the key core parameters of each node in real time and judging the mutation, the window period encryption signal is triggered, the dynamic compression of the load calculation, sampling and robot scheduling period is realized, and the problem of response lag of the traditional fixed period to the sudden fluctuation is solved. For example, when the seawater bromine concentration suddenly drops from 50ppm to 30ppm (mutation occurs), the system identifies the window period encryption signal, shortens the medium load integration period from 20 minutes to 10 minutes, shortens the automatic sampling period from 30 minutes to 15 minutes, and compresses the robot control scheduling plan period from 2 hours to 1 hour, to ensure the rapid response to low concentration raw materials.

[0020] Parameter mutation refers to that the bromide ion concentration deviates from the normal range in a short time, which is the core signal for triggering scheduling optimization; The window period encryption signal dynamically shortens the period, so that the system collects data, calculates load and plans tasks at a higher frequency to adapt to the process demand after the core parameters are adapted.

[0021] In the analysis process of bromine extraction process, the blowing curve is constructed based on historical data, the rapid growth section and the slow saturation section are divided, the real-time load change stage is accurately located by comparing the current load growth slope with the historical stage slope range, and the basis for differentiated scheduling is provided. For example, if the current medium load growth slope is 5 kg / h, which falls within the historical rapid growth section slope range [4, 6] kg / h, it is determined that it is in the rapid growth section, and the robot performs the task according to the high-frequency sampling (15 minutes / time) and linear load prediction model; if the slope decreases to 1 kg / h, which falls within the slow saturation section range [0.5, 2] kg / h, the exponential prediction model is switched to and the medium replacement is planned in advance.

[0022] The historical load characteristic curve is used to record the change law of the blowing efficiency (dynamic load) with time under the same process condition, and provides a benchmark for the current stage judgment; The real-time load change stage is used to distinguish between sufficient (rapid growth section) and insufficient (slow saturation section) active sites of the medium, and to ensure that the scheduling strategy matches the medium performance.

[0023] By comparing the deviation of the current dynamic load state and the predicted value, the periodic adjustment and the robot task optimization are linked to realize the double guarantee of high-frequency monitoring and accurate operation when the concentration fluctuates or the load deviates. For example, if the current load in the rapid growth section is 15% lower than the predicted value (deviation exceeds the standard), the double cooperative adjustment mechanism is started: on the one hand, the period is encrypted (such as 10 minutes / time load calculation), and on the other hand, the instruction flow adjustment robot increases the raw material conveying flow from 20 m³ / h to 25 m³ / h (adjusted by frequency converter), which increases the input of bromine ions; if the current load in the slow saturation section is 20% higher than the predicted value, the medium replacement robot is triggered to enter the preparation state 2 hours in advance to avoid bromine ion leakage.

[0024] The double cooperative adjustment mechanism forms a closed loop control by combining period encryption (improving data timeliness) and robot task adjustment (solving process deviation); The frequency converter adjustment and the medium replacement operation correspond to the core operations of the rapid growth section (optimizing raw material input) and the slow saturation section (ensuring medium state), respectively, to realize targeted optimization.

[0025] In summary, the present application solves the problem that the static period in the traditional scheduling cannot adapt to the dynamic process through the logical chain of parameter mutation monitoring, stage dynamic identification and double cooperative adjustment, makes the industrial robot task scheduling more accurate and efficient. In practical application, it can reduce the risk of medium overload leakage, improve the bromine blowing efficiency, reduce the invalid operation of the robot, and further improve the intelligent level and economic benefit of the bromine extraction process.

[0026] Specifically, when the bromine concentration of the raw seawater (waste heat water from the steel plant) suddenly decreases, continuing to operate according to the original process will result in a far lower blowout efficiency of the blowout tower, causing significant waste; when the bromine concentration increases, the fixed cycle may also cause the blowout tower to leak after being overloaded, so the fixed collection cycle needs to be dynamically adjusted. Embodiment 2 Please refer to Figure 1 Specifically: In the bromine extraction process, multiple node automatic sampling valves and bypass sampling pipes are arranged along the conveying and process main flow, and a conveying pump is used to send the detection sample into a colorimetric analyzer to periodically detect the key core parameters of different process sections, and a thermometer and an electromagnetic flowmeter detection device are used to obtain the volume flow rate at the inlet of the bromine extraction process tower; The detection sample refers to the part of seawater extracted from the different process flow pipelines through the multiple node automatic sampling valves and bypass sampling pipes, which is used to send the liquid sample into the online colorimeter or ion chromatograph for detection, and it represents the actual process fluid at a certain time and at a certain process node. The conveying pump refers to the main pump used in the bromine extraction plant to convey the raw seawater from the water collection tank and water storage tank to the subsequent pretreatment unit, blowout tower, etc.

[0027] The bypass sampling pipe refers to a branch pipeline that is introduced from the process flow pipeline (such as the oxidation pipeline, the blowout tower outlet pipeline, the absorption tower outlet pipeline, etc.) through a valve or a micro-flow regulator. Through this small pipeline, a small amount of sample in the main fluid can be introduced to the online analysis equipment (colorimeter, ion chromatograph, etc.).

[0028] Different process sections refer to different physical / chemical treatment links in the seawater bromine extraction process.

[0029] The key core parameters of different process sections are collected in time series to obtain the key core parameters (including temperature, seawater bromine content, chlorine dosage, sodium bromate and sodium bromide ratio, bromine purity, etc.); Through multi-point concentration detection (in different pipelines and process sections), the concentration stability of the raw material after different homogenization sections and different buffer tanks is confirmed; The relevant parameter data of the blowout tower, absorption tower, distillation tower, etc. are extracted from the relevant parameter data and marked as pretreatment information; The multiple nodes here include but are not limited to the oxidation pipeline outlet, the blowout tower outlet, the absorption tower outlet, and the distillation tower outlet. The oxidation pipeline outlet is located at the outlet pipeline after seawater oxidation; The blowout tower outlet is located at the feed pipeline into the absorption tower; The absorption tower outlet is located at the distillation tower feed pipeline.

[0030] According to the pre-processing information, the changes of each node at adjacent monitoring time points are analyzed to obtain the change rate of the corresponding node, which is obtained by the following formula: ,in, is the rate of change of the i-th node at time t, is the bromide ion concentration of the i-th node at time t, is the i-th node at time The bromide ion concentration at is the time window, t is the moment; Based on the change rate of the corresponding node, it is determined whether the corresponding node has a parameter mutation to identify the window period encryption signal.

[0031] Cycle encryption is to refresh the content that needs to be calculated in order to speed up the process; The key core parameters include bromide ion concentration information obtained in different process sections and at different time periods; Through adaptive statistical methods, based on the change rate of the corresponding node, it is determined whether the corresponding node has a parameter mutation, including: Extract the change rate sequence of the corresponding nodes in the historical data to form a historical sequence; Extract the concentration change rate data of the node for the past several days (e.g., 7 days) and the same period (e.g., 08:00-10:00) from the database to form a historical sequence; Based on the historical series, determine the mean rate of change and the standard deviation of the rate of change; like , it is judged that there is a parameter mutation at the corresponding node, and the scheduling instruction is triggered at this time. Otherwise, it is judged that there is no parameter mutation at the corresponding node, and the scheduling instruction is not triggered at this time; Among them, is the mean rate of change, is the standard deviation of the rate of change, k is a constant, usually ranging from 1 to 3, corresponding to different confidence levels, and the specific value is set by the user (according to actual conditions); It is an abnormal threshold defined based on the confidence level. For example, if k=3, it means that the fluctuation is three times higher than the historical normal fluctuation, which indicates some abnormal event that is not accidental. Determine the timestamps of parameter mutations and mark them as stage points. Based on the stage points, obtain the dynamic loads of the blow-out tower, absorption tower, and distillation tower. Specifically: ,in, is the dynamic load of the medium at time t, M is the mass of the medium currently loaded in the bromine extraction process tower, is the bromide ion concentration at the inlet of the bromine extraction process tower, is the bromide ion concentration at the outlet of the bromine extraction process tower, is the volume flow rate at time t, T represents the time from the start of switching to the current bromine extraction process tower to the stage point; The calculation of the medium dynamic load is to calculate the amount of bromine accumulated and carried by the medium blowing into the tower at this moment.

[0032] In the blowing process, only the medium inside the bromine extraction tower is in direct contact with the flowing seawater and undertakes the task of blowing bromide ions, so when the dynamic load of the medium is indicated as 45 kg of bromine per ton of medium, it actually means that the medium in the current bromine extraction tower has blown out an average of 45 kg of bromine per ton. Therefore, in the integral calculation of the medium dynamic load, the medium refers to the medium inside the blowing tower in the running state, which is used to represent the effective bromine load of the blowing unit in real time and provide a basis for subsequent tower replacement scheduling and regeneration operation.

[0033] In this embodiment, automatic sampling valves and bypass sampling tubes are deployed along the main process of bromine extraction (such as the outlet of the oxidation pipeline, the outlet of the blowing tower, the outlet of the absorption tower, the outlet of the distillation tower, etc.), and the bromide ion concentration is periodically detected by a colorimetric analyzer; at the same time, the volumetric flow rate at the inlet of the blowing tower is collected by an electromagnetic flowmeter, realizing multi-dimensional data collection of concentration and flow rate.

[0034] The multi-node automatic sampling valve further ensures that the key core parameters of different process sections are collected, avoiding the limitations of single-point data; The colorimetric analyzer accurately measures the concentration (accuracy ± 1 ppm) through the color reaction of bromide ions with specific reagents; The electromagnetic flowmeter is a non-contact volumetric flow rate measurement (unit m³ / h), which further avoids the corrosion of high-salt seawater on the equipment.

[0035] In this embodiment, the limitations of traditional single-point detection are broken through, forming a network of key core parameters covering the entire process, and combining flow rate data to provide complete input for subsequent dynamic load calculation, and the data integrity is further improved. For example, in a seawater bromine extraction project, three sampling nodes are set along the outlets of the blowing tower, the absorption tower, and the distillation tower, and sampling is performed every 10 minutes, while the flow rate at the inlet of the blowing tower is monitored in real time by an electromagnetic flowmeter, ensuring the spatio-temporal matching of concentration and flow rate data.

[0036] The bromide ion key core parameters of each node are summarized in time sequence (such as t=0, 5, 10 minutes…), forming a complete set of key core parameters, from which all node data before the inlet of the blowing tower (such as the outlet concentration of the pretreatment unit) are selected and marked as pretreatment information, which is used to analyze the concentration change before the raw material enters the blowing tower.

[0037] Time sequence summary integrates discrete key core parameters along the time axis, facilitating the analysis of concentration fluctuation trends over time; The pre-processing information focuses on the key core parameters before the inlet of the blowout tower, excludes the influence of the blowout process on the raw material itself, and accurately locates the source of parameter mutation (such as raw material itself fluctuation rather than blowout tower failure).

[0038] In this embodiment, time series integration is used to realize time series management of key core parameters, and pre-processing information extraction reduces the range of parameter mutation analysis, improving the efficiency of abnormal positioning. For example, 3 nodes within 1 hour are summarized (12 groups in total), 8 groups of data of seawater into the plant and pre-processing nodes are extracted as pre-processing information, it is found that the concentration drops from 50 ppm to 30 ppm after pre-processing, indicating that the mutation occurs in the pre-processing link rather than in the blowout tower, and through the identified mutation, the next step of scheduling analysis is carried out. Based on the pre-processing information, the concentration change rate (unit: ppm / min) of each node at adjacent time points is calculated by formula, and the fluctuation speed of the concentration is quantified. The change rate reflects the change trend of the concentration with time, which is the core index for judging the mutation; The time window is usually 5-10 minutes, which is used to balance the calculation accuracy and data smoothness, i.e. the concentration difference of adjacent 5 minutes is calculated. The absolute value change of the concentration is converted into the rate change, which is convenient for quantitative evaluation of the severity of the fluctuation and provides a comparable index for subsequent mutation judgment. For example, the concentration of a node at t=10 minutes is 50 ppm, the concentration at t=5 minutes is 45 ppm, =5 minutes, then =(50-45) / 5=1ppm / minute, indicating that the concentration shows a slow upward trend; if the concentration at t=10 minutes increases to 60 ppm, =(60-45) / 5=3ppm / minute, the fluctuation rate increases significantly.

[0039] The historical sequence is used to reflect the change rate range of the node under normal working conditions, which is used as a benchmark for normal fluctuation, and the mean and standard deviation are determined by statistical methods to determine the interval of normal fluctuation (such as k=3, covering 99.7% of normal data); The scheduling instruction is used to inform the system to start window period encryption (such as shortening the sampling period).

[0040] The mutation judgment threshold is adjusted adaptively through historical data to avoid misjudgment of fixed threshold in seasons and raw material changes, and the mutation recognition accuracy is improved to more than 95%. For example, the average change rate of a node in the past 7 days from 08:00 to 10:00 is 1 ppm / min, the standard deviation is 0.5 ppm / min, and the threshold is 1+3×0.5=2.5 ppm / min when k=3. If the current change rate is 3 ppm / min (>2.5), it is determined that there is a parameter mutation, and the scheduling instruction is triggered.

[0041] After determining the timestamp of parameter mutation (phase point), the dynamic load of the medium, i.e., the cumulative amount of bromine blown out by the medium, is calculated by a formula to quantify the total amount of blowing out of the medium in real time, thereby providing a precise load state basis for subsequent blowing-out stage identification and robot task scheduling. Through the progressive logic of data acquisition, preprocessing, change rate calculation, mutation judgment, and load quantification, each step realizes the upgrade from passive monitoring to active perception, thereby providing a reliable basis for dynamic collaborative scheduling of industrial robots and solving the problem of response lag of traditional fixed-cycle scheduling to process fluctuations.

[0042] Embodiment 3 Please refer to Figure 1 Specifically, the encrypted signal of the window period is identified, including: According to the phase point, a scheduling instruction is sent to a centralized scheduling optimization controller, which dynamically obtains an adaptive scaling factor according to the abnormal degree of parameter mutation at the corresponding node, and the adaptive scaling factor is used to quantify the scale of dynamic regulation and control. The centralized scheduling optimization controller is used to dynamically control the medium load integration period, control the automatic sampling period, and control the robot regulation and control scheduling period. The medium load integration period, the automatic sampling period, and the robot regulation and control scheduling period are uniformly compressed by the adaptive scaling factor to obtain a new period at the phase point, and the new period includes a medium load integration new period, an automatic sampling new period, and a robot regulation and control scheduling new period. The new period is used as the encrypted signal of the window period, and the window period encryption state is maintained until it is monitored that there is no parameter mutation at the corresponding node, the window period encryption state is stopped, and the new period is restored to the original period.

[0043] The adaptive scaling factor is obtained by the following formula: wherein, represents otherwise, is the adaptive scaling factor at time t; The new period is obtained by the following formula: wherein, is the new period of the corresponding dynamic control object, is the original period of the corresponding dynamic control object, and x is the number of dynamic control medium load integration period, control automatic sampling period, and control robot regulation and control scheduling period; The centralized scheduling optimization controller is used to uniformly coordinate the dynamic load calculation module, the automatic sampling scheduler, and the robot regulation and control plan scheduler to trigger the encryption mode of the global sampling and scheduling period. The identification of the window period encryption signal actually refers to: when a parameter mutation is detected, a signal is immediately generated to indicate that all subsequent operations of the dynamic load integral operation, concentration sampling period, and robot scheduling of the process tower need to temporarily increase the original normal calculation period (e.g., every 20 minutes) to a higher frequency (e.g., every 10 minutes). The window period encryption state refers to the process of continuously calculating, monitoring, or planning at the new period. The original period refers to the original period of medium load integral, the original period of automatic sampling, and the original period of robot control scheduling. In this embodiment, the stage point of parameter mutation is used as a trigger signal to deliver scheduling instructions to the centralized scheduling optimization controller. The controller dynamically calculates an adaptive scaling factor based on the degree of mutation anomaly (e.g., the magnitude of the concentration change rate exceeding the threshold value) and quantifies the scale of period compression.

[0044] The stage point is the timestamp of the occurrence of parameter mutation (e.g., mutation detected at t=10:00, which is the stage point) and is the time reference for period adjustment.

[0045] The centralized scheduling optimization controller is the core coordination unit responsible for unified control of the periods of load calculation, sampling, and robot scheduling to avoid period confusion of various modules.

[0046] The adaptive scaling factor is a parameter used to determine the compression ratio of the period. The adaptive scaling factor is applied to the original period of medium load integral, automatic sampling, and robot control scheduling, and the new period is calculated by formula to achieve unified dynamic compression of the period.

[0047] The original period is the default basic period of the system, such as the original period of medium load integral being 20 minutes, the original period of automatic sampling being 30 minutes, and the original period of robot control scheduling planning being 3 hours. The new period is the compressed period that dynamically changes with the scaling factor to ensure high-frequency response to mutations.

[0048] The new period is defined as the window period encryption signal, and the system enters the encryption state (runs at the new period). When the parameter mutation disappears, the encryption state is stopped, and the period returns to the original value.

[0049] The window period encryption signal indicates the instruction for the system to enter the high-frequency monitoring / calculation mode with the new period as the core parameter. The window period encryption state is the continuous process of the system running at the new period (e.g., from 10:00 mutation occurrence to 12:00 mutation end, maintaining the encryption state for 2 hours).

[0050] The centralized scheduling optimization controller unifies the coordination of the three compression periods, ensures that the rhythm of load calculation-sampling detection-robot scheduling matches, synchronously updates the key core parameters for calculation and real-time detection results, and avoids load calculation errors caused by period misalignment.

[0051] The dynamic start-stop mechanism of the window period encryption state automatically restores the original period after the mutation ends, reducing the energy consumption of the system running at a high frequency continuously. For example, under certain conditions, mutations occur twice a day for 1 hour each, and the encryption state lasts only 2 hours. Compared with high-frequency operation all day, the robot energy consumption is saved by 15%, and the service life of the analyzer is extended by 20%.

[0052] In summary, this step, through the logical design of hierarchical compression, coordinated regulation, and dynamic start-stop, not only ensures rapid response during parameter mutation, but also avoids resource waste, providing accurate time reference for subsequent medium load calculation and robot task scheduling. It is the core link for the system to achieve dynamic self-adaptation.

[0053] The new period of robot regulation and scheduling refers to the period when the robot plans to change the tower, that is, the time stamp when it needs to change the tower again; Embodiment 4 Please refer to Figure 1 Specifically: based on the complete data in the historical period, a historical load characteristic curve is constructed, and the historical load characteristic curve is used to identify the real-time load change stage corresponding to the current dynamic load state after the step of identifying the real-time load change stage corresponding to the current dynamic load state. Collect complete data on the dynamic load of the medium under the same process conditions as the current process over time during the historical bromine extraction process; According to the characteristics of the bromine extraction process, combined with the complete data, a historical load characteristic curve is constructed through a linear function and an exponential function, and the bromine extraction process is divided into a rapid growth segment and a slow saturation segment; The rapid growth segment is characterized by: as time progresses, the bromine extraction process increases rapidly with the input of bromide ions in the feed, the blowout rate (blowout amount per unit time) is high and relatively stable, and the medium active site is sufficient for bromide ion blowout. The slow saturation segment is characterized by: the bromine extraction process grows gradually, the blowout rate continues to decrease, indicating that the medium active site is gradually occupied, and the blowout is close to saturation.

[0054] During the current dynamic medium load integration process, the dynamic load of the blowout tower, the absorption tower and the distillation tower at the current time is determined; Take the current time as the starting point, and intercept data with the same time span as the historical load characteristic curve (such as selecting the previous 1 hour, the previous 2 hours, etc. Different time windows, and determine the optimal window length according to the process fluctuation characteristics); The current dynamic load state growth slope is calculated, specifically as follows: wherein, is the current dynamic load state growth slope, is the dynamic load of the medium at time is the dynamic load of the medium at time is the length of the intercepted time window (i.e., the time span); The current dynamic load state growth slope is compared with the slope range of each stage in the historical load characteristic curve to determine the real-time load change stage.

[0055] The slope range of each stage in the historical load characteristic curve refers to: under the same medium specification and the same process conditions, in the historical operation data (complete data), the blowout rate (slope) interval corresponding to the rapid growth segment and the slow saturation segment of the dynamic load of the medium changing over time; wherein, the rapid growth segment slope range: when the medium active site is sufficient, the blowout amount grows fast and stably, and in multiple historical periods, the blowout rate (slope) of this stage will be concentrated in a smaller interval (such as [4.5, 5.5] kg / hour); the slow saturation segment slope range: when the medium active site is gradually saturated, the blowout amount slows down, and the blowout rate (slope) of this stage will be concentrated in another smaller interval (such as [0.5, 1.5] kg / hour).

[0056] Specifically, the specific steps for obtaining the slope range are as follows: From the production database, extract complete blowout period data that meet the following conditions: the same medium specification (such as X-7 type anion exchange medium, with a loading capacity of 10 tons); similar process conditions (such as a temperature of about 25°C, a flow rate of about 1.5 m³ / h, and a raw material bromine concentration of about 50 ppm); complete blowout process (from “medium regeneration completion” to “blowout saturation triggering regeneration”).

[0057] Finally, m effective historical period data are obtained, each of which contains a time sequence and a corresponding blowout amount sequence; For each historical period, identify the “rapid growth segment” and the “slow saturation segment”, and for each historical period, extract the slope sequence of the “rapid growth segment” and the “slow saturation segment”; For the rapid growth segment slope sequence and the slow saturation segment slope sequence of the m historical periods, calculate the mean and the standard deviation; Determine the slope range: the rapid growth segment slope range: (the mean in the rapid growth segment slope sequence minus 3, multiplied by the standard deviation in the rapid growth segment slope sequence, the mean in the rapid growth segment slope sequence plus 3, multiplied by the standard deviation in the rapid growth segment slope sequence); Slow saturation segment slope range: (slow saturation segment slope sequence mean - 3, multiplied by slow saturation segment slope sequence standard deviation, slow saturation segment slope sequence mean + 3, multiplied by slow saturation segment slope sequence standard deviation); As production continues, new historical cycle data is added to the historical cycle data periodically (such as every week / month), and steps 1-3 are repeated to recalculate the mean and standard deviation, and dynamically update the slope range of the fast growth segment and the slow saturation segment.

[0058] Take a real scenario of seawater bromine extraction: the raw material bromine concentration is stable at 50ppm, and the flow is 20m³ / h. At this time, the medium has many active sites, and the blowout rate is high, which is determined to be in the fast growth segment. The system will predict the blowout amount according to the linear model, and the robot will sample at a regular frequency (such as every 30 minutes) for 8 hours. After 8 hours, the bromine extraction process is close to 300 kg (assuming that 1 ton of medium saturation is 500 kg), and the blowout rate decreases from 5 kg / hour to 1 kg / hour, which is determined to be in the slow saturation segment. The system switches to the exponential model prediction, and at the same time triggers the robot to take more samples (every 15 minutes), and prepares the medium replacement process in advance.

[0059] The real-time load change stage refers to determining the current stage of the medium in the bromine ion blowout process based on real-time collected seawater bromine extraction process data (such as bromine concentration, flow, temperature, etc.) and comparison analysis with historical data. It can be generally divided into fast growth segment and slow saturation segment; Complete data refers to the data of the dynamic load of the medium changing over time under the same process conditions; The feature of the bromine extraction process is that the initial blowout rate is fast, and as the blowout proceeds, the blowout rate gradually slows down, and when it approaches saturation, the blowout rate tends to zero, so an exponential decay function or a segmented function can be considered for fitting.

[0060] Determining the real-time load change stage includes: If the current dynamic load state growth slope is within the slope range of the fast growth segment, it is determined that the current bromine extraction process is in the fast growth segment; If the current dynamic load state growth slope is within the slope range of the slow saturation segment, it is determined that the current bromine extraction process is in the slow saturation segment; Combine the fast growth segment and the slow saturation segment to form the real-time load change stage.

[0061] Through this segmented comparison, the current blowout stage of the medium can be accurately identified, and key state information can be provided for subsequent process adjustment.

[0062] In this embodiment, based on the historical data of the same medium specification and the same process condition (such as temperature 25±2℃, flow rate 1.5±0.5m³ / h), the blowing-out stage is divided by data fitting, and the historical benchmark is provided for the current blowing-out state.

[0063] The complete data refers to the full cycle load-time data from the new filling of the medium to the blowing-out saturation (such as recording the load every 30 minutes within 12 hours), which ensures that the curve covers the whole blowing-out process.

[0064] In the initial stage of blowing-out (fast growth segment), a linear function can be used for fitting, where Y is the blowing-out amount (dynamic load), X is the time, and k1 and b1 are the fitting parameters; In the late stage of blowing-out (slow saturation segment), an exponential function can be used for fitting, where , and are the fitting parameters.

[0065] Using optimization algorithms such as least squares, parameter estimation is performed on the selected fitting function to minimize the error between the fitting function and the actual data. For the linear function, the least squares method aims to minimize the sum of squared errors, and by taking the partial derivative of k1 and b1 and setting it equal to zero, the optimal estimation values of k1 and b1 are obtained by solving the equation set. For the exponential function, the nonlinear least squares method can be used, and the optimal values of , and are obtained by using iterative algorithms (such as the Levenberg-Marquardt algorithm).

[0066] According to the fitting results, combined with the physical process of the bromine extraction process, the load characteristic curve is divided into different stages, which are divided into fast growth segment and slow saturation segment. In the fast growth segment, the blowing-out amount increases approximately linearly with time, because there are a large number of active sites on the medium surface, which can quickly blow out bromide ions; in the slow saturation segment, the blowing-out amount increases gradually, and tends to a stable value, at this time the active sites on the medium surface are gradually occupied by bromide ions, and the blowing-out rate slows down.

[0067] By distinguishing the current blowing stage, the use of a single model (such as linear fitting) to predict the blowing amount is avoided, which would result in a serious underestimation of the prediction value in the rapid growth stage (as the linear model does not consider the late attenuation) and a serious overestimation of the prediction value in the slow saturation stage (as the linear model does not consider the slowing down of growth). In fact, the essence of stage differentiation is to match different prediction models for different process states: determine the rapid growth stage → call the linear growth model to predict the subsequent blowing amount; determine the slow saturation stage → call the exponential decay model to predict the subsequent blowing amount. In this way, the predicted blowing amount can truly reflect the reasonable blowing trend of the medium at this stage, making the comparison between the actual blowing amount and the predicted value more meaningful, and the deviation is no longer a false deviation caused by the wrong model selection, but a true deviation caused by process abnormalities.

[0068] For example: collect the historical data of 10 batches of the same specification medium, and each batch of data shows that the load increases from 0 to 200 kg in the first 4 hours (slope 50 kg / h, linear characteristic), and then increases from 200 kg to 280 kg in the next 8 hours (slope gradually decreases, exponential decay characteristic), thereby constructing a historical curve: rapid growth stage (0-4 hours, slope 40-60 kg / h), slow saturation stage (4-12 hours, slope 0-10 kg / h).

[0069] Based on the real-time collected concentration and flow rate data, the current medium cumulative load is calculated by integration, and the recent growth slope is calculated by a sliding window to quantify the current blowing rate. The current growth slope is compared with the slope range of each stage of the historical curve to match the corresponding blowing stage and determine the current activity site utilization state of the medium.

[0070] The historical slope range is the historical rapid growth stage slope interval (40-60 kg / h) and the slow saturation stage slope interval (0-10 kg / h), which is used as the threshold standard for stage determination.

[0071] The real-time load change stage is located by slope matching to provide a "state label" (such as adjusting the flow in the rapid stage and preparing to change the tower in the saturation stage) for subsequent robot task scheduling. For example, if the current slope is 50 kg / h, which falls within the historical rapid growth stage slope range (40-60 kg / h), it is determined that the current stage is in the rapid growth stage; if the slope is 8 kg / h, which falls within the slow saturation stage range (0-10 kg / h), it is determined to be in the slow saturation stage.

[0072] The fitting curve based on complete data under the same process conditions avoids the subjectivity of manual experience judgment. For example, the rapid stage slope range (40-60 kg / h) determined by 10 batches of historical data improves the accuracy of stage determination compared to the traditional manually set threshold (30-70 kg / h).

[0073] Real-time load calculation combined with sliding window slope can accurately capture the change in blowout rate. For example, when the concentration of raw materials suddenly drops, causing the slope to drop from 50 kg / h to 30 kg / h, the system can identify the change within 1 window period, which is earlier than the traditional fixed period inspection to find abnormalities.

[0074] After the real-time phase is determined, the robot task can be adjusted accordingly. Specifically, in the rapid growth segment, the flow is adjusted by the frequency converter (e.g., increase the pump frequency by 5% per step when the slope is below 40 kg / h), so that the load growth returns to the normal range; in the slow saturation segment, the medium replacement robot is started in advance (e.g., when the slope drops to 5 kg / h, the disassembly and transportation processes are started in advance), to avoid bromide leakage caused by blowout saturation. In summary, through the logic progression of historical benchmark construction, real-time state quantification, and phase accurate matching, the problems of "blowout phase ambiguity, rate change difficult to capture, and robot scheduling without basis" in the traditional bromine extraction process are solved, providing clear state input for the subsequent dual collaborative adjustment mechanism, and serving as the core bridge connecting process perception and robot action.

[0075] Embodiment 5 Please refer to Figure 1 Specifically, according to the real-time load change phase and combined with the historical load characteristic curve, the deviation of the current dynamic load state is identified to trigger the dual collaborative adjustment mechanism, including: According to the real-time load change phase, the historical dynamic load corresponding to the time stamp on the historical load characteristic curve is determined, and is recorded as the predicted dynamic load; By comparing the predicted dynamic load with the current dynamic load state, the deviation value is determined, specifically: Wherein, is the deviation value, is the predicted dynamic load; If the deviation value exceeds the preset deviation threshold, i.e., a significant deviation from the expected blowout curve is found, the dual collaborative adjustment mechanism is triggered, otherwise, no significant deviation from the expected blowout curve is found, and the dual collaborative adjustment mechanism is not triggered.

[0076] The deviation threshold can be determined according to statistical analysis of historical operation data, such as the maximum deviation of the current stage dynamic load (blowout amount) from the predicted value of the historical curve under normal working conditions, which is used as the threshold; The deviation value is used to identify the deviation of the current dynamic load state; The dual collaborative adjustment mechanism includes: Multi-dimensional period adjustment; if the time corresponding to the current dynamic load state is in the window period encryption state, the window period encryption state is continued, if the time corresponding to the current dynamic load state is not in the window period encryption state, the scheduling instruction is triggered again to realize the window period encryption state; The multi-dimensional cycle adjustment here refers to the start of the window cycle encryption state; Adjust the task scheduling: according to the deviation value (i.e. the deviation of the bromine extraction process stage), adjust the task of the industrial robot in advance, specifically: when the real-time load change stage is the rapid growth segment, the industrial robot is dominated to perform the frequency converter adjustment operation, and when the real-time load change stage is the slow saturation segment, the industrial robot is dominated to start the medium replacement operation in advance or delay; The industrial robot is dominated to start the medium replacement operation in advance or delay, which is used to avoid the leakage of bromide ions due to medium overload, affecting the subsequent process; The specific execution process of the industrial robot dominated to perform the frequency converter adjustment operation: the flow adjustment robot receives the instruction→ moves to the delivery pump control cabinet→ reads the current pump frequency parameter→ adjusts the frequency converter according to the preset step (increases or decreases by 5% each time)→ waits for 30 seconds for the stabilization period→ the pressure sensor feeds back the pipeline pressure→ if the pressure is within the safe range and the flow does not reach the upper limit→ repeat the adjustment until the target flow is reached; wherein, if the current dynamic load state is lower than the predicted dynamic load, increase by 5%, and if the current dynamic load state is not lower than the predicted dynamic load, decrease by 5%; The blowing amount of the rapid growth segment depends on the total amount of bromide ions in the raw material per unit time (= concentration x flow), if the blowing amount grows too slowly, it is usually that the input of bromide ions in the raw material is insufficient (such as sudden decrease in concentration, or low flow), which causes the medium to be not full (active sites are not fully utilized), so at this time the raw material delivery flow needs to be increased; In actual production, the bromine concentration in seawater will fluctuate greatly due to seasonal rainfall, changes in ocean currents, etc. (such as a sudden drop from 50ppm to 30ppm). But the original system still takes samples for detection by the robot according to the preset frequency (for example, once every hour), which causes: within dozens of minutes or even hours after the parameter mutation occurs, the system cannot timely capture the fact that the concentration has decreased. The process continues to be executed according to the high-concentration operation plan, and the medium consumption efficiency is significantly reduced. For example: if the preset hourly equivalent bromine extraction process load is 45 kg of bromine per ton of medium, after the concentration suddenly drops, the actual blowing amount is only 30 kg per ton, which causes the medium to be "out of life" prematurely, and the robot is instructed to replace it, resulting in medium waste. If the concentration suddenly rises, the medium quickly saturates, and the system does not increase the sampling frequency, and cannot timely switch to the desorption process, causing the blowing tower to be overloaded, bromide ions to leak into the waste liquid, and the bromine recovery rate of the subsequent purification unit to be reduced, so at this time the industrial robot is dominated to start the medium replacement operation in advance or delay; Combine multi-dimensional cycle adjustment and notification task scheduling adjustment to obtain a double cooperative adjustment mechanism.

[0077] When the real-time load change stage is the slow saturation segment, the industrial robot is dominated to start the medium replacement operation in advance or delay, including: If the current dynamic load state is higher than the predicted dynamic load, a pre-start instruction is sent to the medium replacement robot cluster to perform the early tower replacement operation; the pre-start instruction is used to prepare for the early tower replacement operation, and the medium replacement robot cluster includes a plurality of groups of industrial robots, including but not limited to a carrying robot, a pipe dismounting robot, a cleaning robot, and a sealing detection robot; The preparation operation includes: the system sends a pre-start instruction to the medium replacement robot cluster → the carrying robot checks the new medium storage position and quantity → the pipe dismounting robot reaches the designated station and prepares the tool → the cleaning robot checks the high-pressure cleaning equipment and the cleaning agent storage capacity → the sealing detection robot calibrates the detection instrument; If the medium is saturated in advance, continuing to pass the raw material will cause the bromine ions to be unable to be blown out and directly leaked (with waste water), resulting in waste of raw materials and pollution of the environment, so the medium replacement robot task needs to be planned in advance (such as preparing 2 hours in advance), to try to complete the medium regeneration or replacement before the leakage occurs, to ensure the bromine recovery rate of the subsequent process.

[0078] If the current dynamic load state is not higher than the predicted dynamic load, the current dynamic load state is integrated according to the multi-dimensional period adjustment, and the time stamp of the delayed start of the medium replacement operation is determined according to the integral calculation result and the maximum load capacity, which refers to the maximum load capacity (i.e., the maximum blowing capacity) that the medium can bear.

[0079] After the time stamp of the delayed start of the medium replacement operation is determined, a medium replacement robot task is generated, which includes: emptying the residual liquid in the blowing tower, dismounting the old medium conveying pipeline, carrying the old medium to the regeneration unit, filling the new medium, restoring the pipeline connection and sealing detection, planning the robot action steps and path according to the medium filling amount and the carrying equipment capacity, to ensure efficient and safe replacement.

[0080] Among them, in the seawater bromine extraction process and industrial robot task scheduling scene, the regeneration unit refers to a special process module for desorption, cleaning, and activation treatment of the saturated old medium to restore its blowing performance, which is the core link of realizing medium recycling.

[0081] In this embodiment, the historical dynamic load corresponding to the time stamp is extracted from the historical load characteristic curve as the predicted dynamic load, and the deviation value is calculated by the formula, when the deviation value exceeds the preset threshold value (such as 1.2 times of the maximum deviation of the historical normal working condition), the double cooperative adjustment mechanism is triggered. This step accurately compares the current dynamic load state with the predicted value of the historical data of the same stage, so that the quantization error of the load offset degree is controlled within 5%, which further improves the accuracy compared with the traditional experience judgment offset method (the error is often 15%-20%). For example, the preset deviation threshold value of the rapid growth segment is 3 kg, if the current load is 45 kg and the predicted load is 50 kg, ΔL=5 kg exceeds the threshold value, the system can accurately identify the insufficient blowing offset state, which can provide the "trigger signal" and "adjustment direction" for the double cooperative adjustment, and ensure that the subsequent robot action is based on the clear offset type (such as too high / low load), avoiding blind adjustment without a clear target.

[0082] If the current is in the window period encryption state, it is maintained, otherwise the scheduling instruction is retriggered, the period is compressed through the adaptive scaling factor, and the medium load integration and automatic sampling synchronous encryption are realized. By dynamically maintaining or starting the window period encryption state, the data analysis frequency of the system is increased by 1-2 times when the load is offset, the data timeliness is enhanced, and more intensive decision basis is provided for task adjustment. For example, when the load deviation of the slow saturation segment exceeds the threshold value, the load calculation is updated every 10 minutes after starting encryption, which is 10 minutes earlier than the original 20-minute period to find the trend of continuously high load, which can help to build a positive cycle of offset-encryption-accurate analysis, and ensure continuous tracking of load abnormalities, avoiding the problem of delayed adjustment due to data update lag; According to the real-time load change stage and the deviation value, the industrial robot is scheduled to perform targeted work-the frequency converter adjusts the flow (±5% step each time) in the rapid growth segment, and the medium is replaced in advance / deferred in the slow saturation segment (such as starting the robot cluster to prepare 2 hours in advance when the load is too high). The beneficial effect is that the robot task is accurately matched with the load offset characteristics, and the adjustment efficiency is improved by more than 40%. For example, when the load of the rapid growth segment is 5 kg lower than the predicted value, the flow adjustment robot increases the pump frequency by 5% every 30 seconds, and after 3 adjustments, the load growth rate can return to the expected value; when the load of the slow saturation segment is 8 kg higher than the predicted value, the medium replacement robot starts to disassemble the tool for inspection in advance, which is 1.5 hours earlier than the original plan to complete the tower replacement, avoiding bromine ion leakage. Its role can convert the abstract load offset data into specific actions of the robot, realize the closed loop from data perception to physical intervention, directly solve the problem of efficiency decline or risk increase caused by load offset, and ensure the stability and economy of the bromine extraction process.

[0083] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A method for collaborative scheduling of industrial robot tasks for smart factories, characterized by: The following steps are included: In the bromine extraction process, key core parameters are obtained. Based on the key core parameters, it is determined whether parameter mutations occur at each node along the transportation and main process flow to identify window cycle encrypted signals. The window cycle encrypted signals are used to determine the new cycle of medium load integration, automatic sampling, and robot control and scheduling. The key core parameters include but are not limited to temperature, seawater bromine content, chlorine dosage, sodium bromate and sodium bromide ratio, and bromine purity. Based on the complete data within the historical period, a historical load characteristic curve is constructed. The historical load characteristic curve is used to identify the real-time load change stage corresponding to the current dynamic load state. The real-time load change stage includes a rapid growth stage and a slow saturation stage. Based on the real-time load change stage and combined with the historical load characteristic curve, the offset degree of the current dynamic load state is identified to trigger the dual collaborative adjustment mechanism. The dual collaborative adjustment mechanism is used to adjust the inverter and identify the startup medium replacement operation through the industrial robot.

2. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 1, characterized in that: In the bromine extraction process, multi-node automatic sampling valves and bypass sampling tubes are installed along the transportation and main process flow. The test samples are sent to the colorimetric analyzer through the delivery pump to periodically test the key core parameters of different process sections. The volume flow rate at the inlet of the bromine extraction process tower is obtained through the use of thermometers and electromagnetic flowmeters. Summarize the key core parameters of different process sections according to time series to obtain the key core parameters; Extract parameter data on the nodes of the blow tower, absorption tower, and distillation tower and mark them as preprocessing information; According to the pre-processed information, the changes of each node at adjacent monitoring time points are analyzed to obtain the change rate of the corresponding node; Based on the change rate of the corresponding node, it is determined whether the corresponding node has a parameter mutation to identify the window period encryption signal.

3. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 2, characterized in that: Based on the change rate of the corresponding node, determine whether the corresponding node has a parameter mutation, including: Extract the change rate sequence of the corresponding nodes in the historical data to form a historical sequence; Based on the historical series, determine the mean rate of change and the standard deviation of the rate of change; like , it is judged that there is a parameter mutation at the corresponding node, and the scheduling instruction is triggered at this time. Otherwise, it is judged that there is no parameter mutation at the corresponding node, and the scheduling instruction is not triggered at this time; Among them, is the rate of change of the i-th node at time t, is the mean rate of change, is the standard deviation of the rate of change, k is a constant; The timestamps of parameter mutations are determined and marked as stage points. Based on the stage points, the dynamic loads of the blow-out tower, absorption tower, and distillation tower are obtained.

4. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 3, characterized in that: Identify window period encryption signals, including: According to the stage point, the scheduling instruction is sent to the centralized scheduling optimization controller. The centralized scheduling optimization controller will dynamically obtain the adaptive scaling factor according to the abnormal degree of parameter mutation in the corresponding node. The adaptive scaling factor is used to quantify the scale of dynamic control. The centralized scheduling optimization controller is used to dynamically control the medium load integration period, the automatic sampling period, and the robot control scheduling period. The medium load integration period, the automatic sampling period, and the robot control and scheduling period are uniformly compressed by an adaptive scaling factor to obtain a new period at a stage point. The new period includes a new medium load integration period, a new automatic sampling period, and a new robot control and scheduling period. The new cycle is used as the window cycle encryption signal, and the window cycle encryption state is maintained until it is detected that there is no parameter mutation in the corresponding node, the window cycle encryption state is stopped, and the new cycle is restored to the original cycle.

5. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 4, characterized in that: After constructing a historical load characteristic curve based on complete data within a historical period, and using the historical load characteristic curve to identify the real-time load change stage corresponding to the current dynamic load state, the following steps are also included: Collect complete data on the dynamic load of the medium over time during the historical bromine extraction process under the same process conditions as the current one; According to the characteristics of the bromine extraction process, combined with complete data, a historical load characteristic curve is constructed through linear and exponential functions, and the bromine extraction process is divided into a rapid growth stage and a slow saturation stage; During the current dynamic medium load integral accumulation process, the dynamic loads of the blow-out tower, absorption tower, and distillation tower at the current moment are determined; Record the current moment as the starting point and intercept the data with the same time span as the historical load characteristic curve; Calculate the growth slope of the current dynamic load state; The current dynamic load state growth slope is compared with the slope range of each stage in the historical load characteristic curve to determine the real-time load change stage.

6. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 5, characterized in that: Determine the real-time load change phase, including: If the current dynamic load state growth slope is within the slope range of the rapid growth stage, it is determined that the current bromine extraction process is in the rapid growth stage; If the current dynamic load state growth slope is within the slope range of the slow saturation stage, it is determined that the current bromine extraction process is in the slow saturation stage; The fast growth segment and the slow saturation segment are combined to form a real-time load change stage.

7. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 6, characterized in that: Based on the real-time load change stage and combined with the historical load characteristic curve, the deviation degree of the current dynamic load state is identified to trigger the dual coordinated adjustment mechanism, including: According to the real-time load change stage, the historical dynamic load at the corresponding timestamp on the historical load characteristic curve of the current dynamic load state is determined and recorded as the predicted dynamic load; By comparing the predicted dynamic load with the current dynamic load state, a deviation value is determined; If the deviation value exceeds the preset deviation threshold, the dual collaborative adjustment mechanism is triggered.

8. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 7, characterized in that: Dual coordinated adjustment mechanism, including: Multi-dimensional cycle adjustment: If the moment corresponding to the current dynamic load state is in the window cycle encryption state, the window cycle encryption state will continue to be maintained. If the moment corresponding to the current dynamic load state is not in the window cycle encryption state, the scheduling instruction will be triggered again to achieve the window cycle encryption state. Notify task scheduling adjustments: Adjust the tasks of the industrial robot in advance based on the deviation value. Specifically, when the real-time load change phase is a rapid growth phase, the industrial robot is instructed to perform inverter adjustment operations. When the real-time load change phase is a slow saturation phase, the industrial robot is instructed to advance or postpone the start of the medium replacement operation. Combine multi-dimensional period adjustment and notification task scheduling adjustment to obtain a dual collaborative adjustment mechanism.

9. The method for collaborative scheduling of industrial robot tasks for a smart factory according to claim 8, characterized in that: When the real-time load change phase is a slow saturation phase, the industrial robot is instructed to start the medium replacement operation in advance or in a delayed manner, including: If the current dynamic load state is higher than the predicted dynamic load, a pre-start instruction is sent to the medium replacement robot cluster. The pre-start instruction is used to prepare for the tower replacement operation in advance. The medium replacement robot cluster includes several groups of industrial robots. If the current dynamic load state is not higher than the predicted dynamic load, the current dynamic load state is promptly integrated according to the multi-dimensional period adjustment. The timestamp for delaying the start of the media replacement operation is determined based on the integral calculation result and the maximum load. The maximum load refers to the maximum load that the medium can bear.

Images