A three-product heavy medium cyclone re-separation system for middlings in jigging

Abstract

The application relates to the field of automatic control of a coal preparation process, and discloses a three-product heavy medium cyclone re-cleaning system for middlings in a jig, which comprises a data acquisition module, a flow state decoupling extraction module, a state determination module, a steady-state adjustment module and an intervention interlocking module; the data acquisition module synchronously acquires instantaneous physical quantities of an equipment layer, carries out translation compensation based on physical material conveying time, acquires time sequence aligned synchronous jig variance, the flow state decoupling extraction module carries out reference frequency normalization compensation on electrical parameters according to a fluid machine similarity law, extracts and removes a normalized flow rheological energy ratio after frequency interference, the state determination module cross-compare the above two parameters with preset threshold values and outputs a control instruction, according to the instruction, the steady-state adjustment module executes variable parameter steady-state control, or the intervention interlocking module cuts off a conventional closed loop and executes a physical pollution cleaning action. The application eliminates time lag and frequency speed regulation interference, avoids adjustment misjudgment of a washing and cleaning system and equipment jam.

Description

Technical Field

[0001] This invention relates to the field of home kitchen technology, specifically to a jigging, pressureless three-product heavy medium cyclone reselection system for coal. Background Technology

[0002] In coal washing processes, pressureless three-product heavy medium cyclones are commonly used for the re-selection of middlings from jigging. The feed pump, as the core power unit of the re-selection system, directly determines the separation performance of the heavy medium cyclone. Typically, variable frequency speed control technology is used on-site to control the feed pump, and this is coordinated with the operating conditions of the upstream jig.

[0003] However, existing control systems have limitations in handling complex operating conditions. There is a physical spatial span between the upstream jig and the downstream feed pump, and the material transport in pipelines and chutes requires a certain amount of time. Existing systems often directly collect and compare instantaneous data from both ends of the equipment without considering the lag time of physical transport, causing a misalignment between the operating condition judgment signal and the actual material arrival status on the time axis.

[0004] Meanwhile, when the feed pump undergoes variable frequency speed regulation, the electrical parameters of the motor will change nonlinearly. Existing technologies often directly use measured active power or current as the basis for determining the fluid rheological state. Because they fail to eliminate the interference from the electrical parameter changes caused by the variable frequency speed regulation itself, the monitoring data cannot accurately reflect the actual rheological conditions within the pipeline. When fluctuations in the jig's operating conditions occur simultaneously with the feed pump's variable frequency adjustment, the time misalignment and parameter distortion combine, making the original single-parameter closed-loop control logic prone to outputting incorrect adjustment commands. This misjudgment not only makes it difficult to stabilize the washing system but also easily causes control loop oscillations, leading to pipeline blockage and jamming of underlying equipment, affecting the continuous operation of the production process. Summary of the Invention

[0005] The technical problem solved by this invention is that in the jigging coal re-selection process, due to the time lag in the physical conveying of materials and the change in the operating conditions caused by the variable frequency speed regulation of the feed pump, it is difficult to obtain accurate fluid rheological state. Single parameter monitoring is prone to causing the washing and beneficiation system to make incorrect judgments under complex operating conditions, which in turn leads to the failure of conventional closed-loop regulation or physical jamming of equipment.

[0006] To address the above problems, the present invention provides the following technical solution:

[0007] This invention provides a jigging, pressureless three-product heavy medium cyclone reselection system for middlings coal, the system comprising:

[0008] The data acquisition module is used to synchronously acquire the instantaneous physical quantities of the jig and feed pump equipment layers, and perform time dimension translation compensation based on the inherent physical conveying time of the material to obtain the synchronous jig variance after time alignment.

[0009] The rheological state decoupling extraction module is used to perform reference frequency normalization compensation on the actual electrical parameters based on the instantaneous physical quantities of the feed pump and the fluid mechanical similarity law, and extract the normalized rheological energy ratio after eliminating frequency conversion interference.

[0010] The state determination module is used to compare the synchronous jigging variance with the preset jigging disorder variance threshold and the normalized rheological energy ratio with the preset rheological energy ratio threshold. Based on the cross-comparison results of the two-dimensional state matrix, it outputs the operating condition routing control command.

[0011] The steady-state adjustment module is used to perform variable parameter steady-state control by using the normalized rheological energy ratio as a feedforward factor when the operating condition routing control command is a normal adjustment signal; the intervention interlock module is used to cut off the normal closed-loop circuit and perform time-interlocked physical cleaning protection actions when the operating condition routing control command is an intervention trigger signal.

[0012] The data acquisition module is specifically used to: acquire the instantaneous displacement signal of the upstream jig buoy sensor and calculate the real-time jig variance within a set sliding time window when acquiring the synchronous jig variance after time alignment;

[0013] The ratio of the pre-calibrated physical material transport time to the set high-frequency sampling period is used to calculate the queue depth of the first-in-first-out queue.

[0014] The real-time jigging variance is continuously stored in the input of the first-in-first-out queue, and the jigging variance after periodic delay is extracted from the output of the first-in-first-out queue and used as the synchronous jigging variance after time alignment.

[0015] The rheological state decoupling extraction module is specifically used for extracting the normalized rheological energy ratio after eliminating frequency conversion interference when:

[0016] The instantaneous active power and instantaneous actual output frequency of the feed pump frequency converter, as well as the instantaneous volumetric flow rate of the electromagnetic flow meter in the feed pipeline, are obtained.

[0017] The rated power frequency of the feed pump drive motor is used as the reference frequency. The ratio of instantaneous active power to instantaneous volumetric flow rate is calculated. This ratio is then multiplied by the square of the ratio of the reference frequency and the instantaneous actual output frequency to obtain the normalized rheological energy ratio. When the instantaneous volumetric flow rate or the instantaneous actual output frequency is lower than the set start-stop lower limit threshold, the current normalized rheological energy ratio is automatically locked to the calculated value of the previous effective cycle.

[0018] The state determination module is internally configured with a stabilization time window for filtering out high-frequency transient interference, and the stabilization time window is logically evaluated.

[0019] When the synchronous jigging variance is less than the jigging disorder variance threshold, or the normalized rheological energy ratio is less than the rheological energy ratio threshold, a normalized adjustment signal is output. When the synchronous jigging variance is greater than or equal to the jigging disorder variance threshold, and the normalized rheological energy ratio is greater than or equal to the rheological energy ratio threshold, and the duration of exceeding the limit exceeds the anti-shaking time window, the washing system is determined to be in a complex and severe working condition, and an intervention trigger signal is output.

[0020] The steady-state adjustment module is specifically used for: when performing variable parameter steady-state control.

[0021] The attenuation term is constructed by using the ratio of the normalized rheological energy ratio to the rheological energy ratio threshold, and the basic proportional gain is dynamically corrected to obtain the real-time proportional gain.

[0022] When the real-time proportional gain is lower than the set safety lower limit, the proportional gain of the current cycle is forcibly locked to the safety lower limit, and the accumulation calculation of the internal controller integral term is simultaneously suspended.

[0023] The intervention interlock module is specifically used to: issue opening commands to the diversion valve and clean water supply valve of the washing pipeline and simultaneously read the physical position feedback signal of the corresponding valve when performing the time-interlocked physical cleaning protection action;

[0024] If the target opening signal is not detected within the set time limit, a hardware fault interruption alarm will be triggered, and the upstream main feed gate will be shut down in conjunction with the alarm.

[0025] When the synchronous jigging variance and the normalized rheological energy ratio both fall below the set recovery threshold, and the duration of this satisfactory state exceeds the set reset anti-shake window, the intervention trigger signal is cancelled.

[0026] At the same time, the current actual physical output feedback value of the underlying actuator is assigned in reverse to the integral term of the steady-state control module as the initial starting point to perform a disturbance-free switching.

[0027] Preferably, the system further includes a health assessment and threshold optimization module, used to extract equipment operation data and construct a multi-dimensional input feature vector within a set assessment period;

[0028] The multidimensional input feature vector is fed into a backpropagation neural network containing hidden layers to perform forward inference, calculate and limit the output of the equipment wear attenuation coefficient that characterizes the deterioration of the conveying efficiency;

[0029] Based on the equipment wear attenuation coefficient, the current rheological energy ratio threshold is dynamically lowered.

[0030] In this invention, the system calculates the jigging variance using the inherent physical transport time of the material and performs first-in-first-out queue delay calculation, which compensates for the acquisition time difference caused by spatial distance and realizes the correspondence between the state of upstream equipment and the state of the current equipment on the time axis; the system uses the fluid machinery similarity law to process instantaneous power and flow data, eliminates electrical parameter interference caused by frequency converter speed regulation, and obtains the normalized rheological energy ratio that reflects the rheological characteristics of the pipeline.

[0031] Based on this, the system uses a two-dimensional state matrix and anti-shake time window to make cross-judgments, distinguish between normal working conditions and complex severe working conditions, and select to execute the corresponding actions, thereby avoiding oscillations in the control loop under complex working conditions and ensuring the operational stability of the washing and screening process.

[0032] This invention provides a pressureless three-product heavy medium cyclone reselection system for jigged middlings coal. It has the following advantages:

[0033] 1. This invention obtains the inherent physical conveying time of the material and performs a first-in-first-out queue delay translation compensation for the real-time jigging variance of the upstream jig. This feature compensates for the signal time difference caused by the spatial distance between the equipment, so that the upstream jigging state and the operating state of the feed pump of this stage are aligned on the time axis, avoiding misjudgment of the working condition caused by signal time misalignment.

[0034] 2. Based on the fluid machinery similarity law, this invention uses a reference frequency to perform normalized compensation calculations on the instantaneous active power and volumetric flow rate of the feed pump. This feature eliminates the influence of frequency fluctuations of the frequency converter on electrical parameters, obtains the normalized rheological energy ratio that reflects the true rheological characteristics inside the pipeline, and solves the problem of conventional closed-loop regulation failure caused by frequency converter interference.

[0035] 3. This invention utilizes a two-dimensional state matrix to cross-compare the synchronous jigging variance and the normalized rheological energy ratio, and outputs corresponding control commands based on the results. Under normal operating conditions, the system uses the rheological energy ratio as a feedforward factor to perform variable parameter steady-state control. Under complex and severe operating conditions, the system disconnects the conventional closed-loop circuit and performs time-interlocked physical cleaning actions to prevent control loop oscillation and physical jamming of equipment. Attached Figure Description

[0036] Figure 1 This is a flowchart of the control method for the heavy medium cyclone reselection system according to an embodiment of the present invention;

[0037] Figure 2 This is a schematic diagram illustrating the data acquisition and timing alignment processing principle of an embodiment of the present invention.

[0038] Figure 3 This is a schematic diagram illustrating the principle of rheological state decoupling and extraction in an embodiment of the present invention.

[0039] Figure 4This is a two-dimensional state matrix determination and routing logic diagram according to an embodiment of the present invention;

[0040] Figure 5 This is a schematic diagram illustrating the adaptive control and protection execution principle of an embodiment of the present invention;

[0041] Figure 6 This is a schematic diagram illustrating the principle of device health assessment and dynamic threshold optimization in an embodiment of the present invention.

[0042] Figure 7 This is a system hardware deployment and communication topology diagram according to an embodiment of the present invention;

[0043] Figure 8 The simulation curves show the comparison of the control strategy effects in embodiments of the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Example:

[0046] See attached document Figure 1 , Figure 1 This is a flowchart of a control method for a jigging, pressureless three-product heavy medium cyclone reselection system for coal according to an embodiment of the present invention.

[0047] This invention provides a control method for a jigging, pressureless three-product heavy medium cyclone reselection system for coal, comprising the following steps:

[0048] S10: Acquire the instantaneous physical quantities of each device in the system according to the set high-frequency sampling period, establish a first-in-first-out queue to perform spatial hysteresis compensation on the variance of the upstream jig buoy displacement data, and output the synchronous jig variance.

[0049] S20: Obtain the electrical parameters of the frequency converter of the hydrocyclone feed pump and the flow data of the feed pipe. Use the fluid machinery similarity law to eliminate the nonlinear interference of the frequency converter output frequency change on the power and flow ratio, and calculate the normalized rheological energy ratio.

[0050] S30 substitutes the synchronous jigging variance and the normalized rheological energy ratio into the preset two-dimensional state matrix for judgment. Based on whether the two data exceed the corresponding threshold at the same time, it outputs a normal adjustment signal or an intervention trigger signal.

[0051] S40, when receiving a regular adjustment signal, performs density feedforward compensation and pressure dynamic compensation in parallel, adjusts the opening of the qualified medium tank water supply regulating valve using the real-time jigging variance without delay processing, and updates the frequency setting value of the feed pump frequency converter according to the change in the normalized rheological energy ratio.

[0052] S50, when receiving the intervention trigger signal, freezes the current control command of the water supply regulating valve and the feed pump frequency converter, and outputs a drive signal to the desliming diversion regulating valve according to the integral accumulation of the normalized rheological energy ratio exceeding the threshold. When the normalized rheological energy ratio falls back to the preset safe range, the freezing state is lifted.

[0053] S60 obtains the actual clean coal ash content output by the downstream online ash analyzer according to the set low-frequency sampling period, calculates the deviation between the actual ash content and the target ash content, and uses a recursive least squares algorithm with a forgetting factor to periodically correct the basic density setpoint, basic pressure setpoint, and reference energy ratio.

[0054] See attached document Figure 2 , Figure 2 This is a schematic diagram of data acquisition and timing alignment processing according to an embodiment of the present invention.

[0055] The data acquisition module provided by this invention is responsible for communicating with the underlying field equipment to complete the synchronous acquisition of multi-source operational status data and the alignment and mapping of the time dimension. In this embodiment, the data acquisition module specifically executes the following sub-steps to complete the processing logic of relevant parameters:

[0056] S101, the control system operates according to the set high-frequency sampling period. Execute data read commands to synchronously acquire various instantaneous physical quantities from the device layer;

[0057] Specifically, the acquired data includes the instantaneous displacement signal from the upstream jig buoy sensor. Instantaneous active power of the frequency converter of the hydrocyclone feed pump The instantaneous actual output frequency of the feed pump inverter And the instantaneous volumetric flow rate of the electromagnetic flowmeter in the feed pipeline. High-frequency sampling period The set values ​​must satisfy the sampling theorem.

[0058] As a preferred method, high-frequency sampling period The interval is typically set to 0.5 to 2 seconds to facilitate the capture of sudden changes in fluid state and avoid consuming too much system computing power.

[0059] For the specific hardware and software configuration of the programmable logic controller to read the physical analog quantities of the underlying sensors and the electrical parameters of the internal registers of the frequency converter through the industrial communication bus, those skilled in the art can make conventional settings according to the on-site hardware selection and network architecture. The communication protocol configuration and hardware address mapping are well-known technologies in this field and will not be described in detail here.

[0060] S102. After completing the collection of basic data, the system needs to further extract operating characteristic parameters that can characterize process fluctuations.

[0061] The stability of the stratification state of the bed inside the jig is related to the properties of the discharged middlings material.

[0062] Under normal circumstances, when bed instability causes high-frequency and large-amplitude fluctuations in buoy displacement, it means that the content of extremely difficult-to-sort particles with a density close to the sorting density in the discharged middlings may suddenly increase.

[0063] To this end, the data acquisition module allocates a data buffer area in the memory space of the control system, and establishes a buffer containing... Sliding time window for data at each sampling point.

[0064] The control system calculates the instantaneous displacement signal of the buoy within the current time window. Real-time jigging variance The real-time jigging variance The specific calculation formula is as follows:

[0065] In the formula, Indicates the sampling time The corresponding real-time jigging variance; This represents the total number of discrete sampling points contained within the sliding time window. To avoid calculating dead zones and ensure statistical significance, this value must be greater than 1.

[0066] As a specific implementation scenario, The pulsation cycle of the jig can be set to 60 to 120 sampling points, corresponding to a time span of approximately 1 to 2 minutes;

[0067] This indicates counting backwards from the current sampling time. The instantaneous displacement signal value of the buoy in each sampling period;

[0068] It represents the arithmetic mean of all instantaneous displacement signals within the current sliding time window.

[0069] S103, after obtaining the real-time jigging variance, the system faces the technical problem of how to match the upstream data with the downstream data;

[0070] From a physical transport perspective, the middlings material discharged from the jig passes through equipment such as dewatering screens, conveyor belts, and chutes to reach the downstream hydrocyclone feed pump. The entire process has an inherent physical transport time. If time alignment is not performed, directly comparing the current jig state with the current pump fluid state will cause the judgment logic to apply to different batches of materials.

[0071] To ensure that the upstream material state parameters are aligned and verified with the rheological parameters of the batch of material when it actually arrives at the feed pump in the time dimension, the data acquisition module establishes a depth of [missing information] within the control system. A first-in, first-out (FIFO) queue. The formula for calculating the depth of this queue is: Among them, the physical conveying time of materials The value can be obtained by dividing the physical length of the on-site conveyor belt by the rated belt speed and adding the free fall time of the material in the chute for preliminary calculation and calibration.

[0072] The control system will use the real-time jigging variance calculated in step S102. The data is continuously stored at the input of the FIFO queue and extracted from the output of the FIFO queue after a fixed period delay. This data is defined as the synchronous jigging variance. synchronous jigging variance With real-time jigging variance The relation is:

[0073] ;

[0074] In the formula, This represents the synchronous jigging variance that corresponds to the fluid in the current feed pump in terms of time sequence. This indicates the aforementioned physical material transport time.

[0075] By shifting and accessing data between register addresses at a specific depth, the data shift compensation in the time stream is completed, and the output synchronous jigging variance is calculated. It can be directly used as the input condition for determining the subsequent state matrix.

[0076] See attached document Figure 3 , Figure 3 This is a schematic diagram of the rheological state decoupling extraction principle according to an embodiment of the present invention.

[0077] The rheological state decoupling extraction module provided by this invention receives operating data transmitted from the underlying device, which is used to reduce the impact of changes in device operating status on fluid characteristic evaluation and output characteristic parameters characterizing the physical properties of suspension.

[0078] In the pressureless three-product heavy medium cyclone washing process, when the feed contains fine coal slime, the apparent viscosity of the heavy suspension will increase accordingly.

[0079] High-viscosity fluids need to overcome more frictional resistance when transported in pipelines. In this embodiment, the rheological state decoupling extraction module establishes the mapping relationship between equipment electrical parameters and fluid viscosity and performs decoupling calculations through the following sub-steps:

[0080] S201, Establish the initial mapping relationship of rheological characteristics;

[0081] The control system retrieves the instantaneous active power of the feed pump inverter. and the instantaneous volumetric flow rate of the electromagnetic flowmeter in the feed pipeline From a physical mechanism perspective, under constant speed conditions, the amount of active power consumed by a centrifugal pump to maintain the same flow rate can reflect the changing trends of internal friction and apparent viscosity of the fluid in the pipeline network. Based on this, the control system initially extracts the ratio of active power to volumetric flow rate, which is used as the basic data for assessing the degree of high mud pollution in the fluid.

[0082] After obtaining the initial data, the interference of variables caused by the adjustment of the equipment's own operating conditions must be considered. In actual industrial sites, the feed pump is usually in a dynamic variable frequency operation state in order to adapt to the hydrocyclone inlet pressure adjustment requirements under different operating conditions.

[0083] Changes in the inverter's output frequency can have a nonlinear effect on the initial ratio, leading to deviations in the extraction of the true fluid state from the coupled data. To address this data coupling problem, the system needs to further incorporate similarity laws to perform feature compensation.

[0084] S202, performs interference removal and parameter normalization based on the fluid machinery similarity law;

[0085] The operating characteristics of a centrifugal pump follow the similarity law: its active power consumption is proportional to the cube of the frequency converter output frequency corresponding to the equipment speed, while its output volumetric flow rate is proportional to the first power of the output frequency.

[0086] It can be deduced that the ratio of active power to volumetric flow rate is directly proportional to the square of the output frequency.

[0087] If the control system directly uses the ratio of instantaneous active power to volumetric flow rate to determine the operating condition, this ratio will drift quadratically with the frequency modulation action, regardless of the actual viscosity of the fluid.

[0088] This kind of undecoupled physical quantity drift can easily be misjudged by the control logic as a deterioration of material properties, which in turn can cause control oscillations in the closed loop.

[0089] To eliminate the physical quantity coupling effects caused by mechanical speed regulation, the rheological state decoupling extraction module introduces a fixed reference frequency. The current electrical parameter ratios are normalized and compensated, and the control system calculates the normalized rheological energy ratio. Its mathematical expression is:

[0090] ;

[0091] In the formula, This represents the normalized rheological energy ratio reflecting the apparent viscosity of the fluid after eliminating the influence of variable frequency speed. This indicates the instantaneous active power of the feed pump frequency converter. This indicates the instantaneous volumetric flow rate of the feed pipe. This indicates the instantaneous actual output frequency of the feed pump inverter. This indicates the set reference frequency.

[0092] In this embodiment, the reference frequency Set the frequency to the rated operating frequency of the feed pump drive motor, for example, 50Hz.

[0093] Meanwhile, to prevent dead zone faults where the denominator of the algorithm is zero during equipment downtime or sensor failure.

[0094] As a preferred approach, the control system sets logical conditions before executing the formula calculation, based on the instantaneous volumetric flow rate. or instantaneous actual output frequency When the device starts or stops below the set threshold, the normalized rheological energy ratio is automatically locked. The calculated value is based on the previous effective period. In order to provide more specific engineering implementation references for those skilled in the art, the aforementioned equipment start-up and shutdown lower limit threshold can usually be set to 5% to 10% of the equipment's rated operating parameters to effectively filter out invalid noise data during the shutdown phase.

[0095] It can be configured in conjunction with the water pump product manual and conventional PLC programming specifications. Its register holding and dead-time limiting logic are well-known technologies in this field and will not be elaborated here.

[0096] Based on the above computational logic, the output normalized rheological energy ratio The characteristic parameter varies with the content of fine coal slime inside the heavy suspension. In mathematical processing, this parameter basically eliminates the interference caused by the current frequency regulation operation of the water pump equipment, thus providing a unified fluid physical parameter for the subsequent control system execution status determination.

[0097] See attached document Figure 4 , Figure 4 This is a two-dimensional state matrix determination and routing logic diagram according to an embodiment of the present invention.

[0098] The status determination module provided by this invention receives the feature parameters processed by the preceding module, combines them with the safety boundary preset by the process requirements, determines the current operating condition of the system, and issues the corresponding routing control command.

[0099] In this embodiment, the state determination module specifically performs the following sub-steps to complete the logic evaluation and signal distribution:

[0100] S301, Set the reference threshold for state determination;

[0101] In the pressureless three-product heavy medium cyclone washing process, fluctuations in a single parameter are insufficient to fully reflect the true operating status of the system.

[0102] To establish a multi-dimensional evaluation system, the control system internally presets a jigging disorder variance threshold. Rheological energy ratio threshold .

[0103] Jigging disorder variance threshold The critical physical point that characterizes the upstream jig machine bed layer as being in a state of normal stratification and disorder.

[0104] As a preferred method, this threshold can be calibrated offline by collecting the statistical upper limit of the buoy displacement variance under historical normal operating conditions of the jig, and adding a preset safety margin coefficient, for example, taking 1.2 to 1.5 times the historical normal average as a reference benchmark. Rheological energy ratio threshold. Characterizes the viscosity carrying capacity of the heavy suspension inside the hydrocyclone.

[0105] When the energy ratio corresponding to the fluid viscosity exceeds the limit, fine coal slime in the centrifugal force field is prone to cause blockage of the underflow port or deviation in the sorting density.

[0106] To provide an implementable engineering reference for those skilled in the art, this rheological energy ratio threshold It can be calibrated as 1.8 to 2.5 times the basic energy ratio of the equipment under clean water operation conditions. The specific value can be finely adjusted and determined in combination with the actual geometric dimensions of the hydrocyclone and the rated load capacity of the feed pump drive motor.

[0107] After establishing the judgment criteria, the control system then performs a comprehensive comparison and analysis on the concurrent real-time data streams.

[0108] S302, construct a two-dimensional state matrix and perform spatiotemporal cross-validation;

[0109] The control system reads the synchronized jigging variance in parallel, aligned with the time dimension, within the same time slice. and normalized rheological energy ratio Since the upstream material state transfer and the downstream fluid evolution are mutually coupled physical processes, the system is constructed with a two-dimensional logic matrix with synchronous jigging variance and normalized rheological energy ratio as input variables.

[0110] The control system compares the two real-time data points with their corresponding preset thresholds. This cross-comparison, based on spatial hysteresis compensation and within the same time coordinate system, helps reduce the action lag and regulation conflicts that often occur in distributed control logic.

[0111] Based on the comparison results of the above matrices, the control system needs to distribute matching underlying control commands.

[0112] S303 triggers the operating condition routing and outputs the underlying control signal;

[0113] The control system performs logical branching based on the cross-comparison results of the two-dimensional state matrix. In this embodiment, the system introduces Boolean logic state variables. The logical expression for determining the switching of operating conditions is as follows:

[0114] ;

[0115] In the formula, This indicates the logic state quantity that triggers the intervention. This represents the variance of synchronous jigging. This represents the threshold of jigging disorder variance. Represents the normalized rheological energy ratio. Indicates the rheological energy ratio threshold. Represents the logical AND operation.

[0116] In industrial environments, sensor data is often accompanied by high-frequency noise. If state switching is based solely on transient value exceeding limits, it can easily lead to frequent state jumps in the control system.

[0117] To ensure the integrity of the algorithm logic and avoid damage to the actuator, as a preferred approach, the control system introduces a debouncing time window in the Boolean decision. It is used to filter out interfering spike signals.

[0118] Specifically, this image stabilization time window The sampling period can be set to 3 to 5 times, depending on the response characteristics of the field equipment.

[0119] During this time window, if the Boolean logic state quantity If the value drops and changes abruptly, the system will reset the timing sequence to ensure that intervention is only responded to persistent deterioration of operating conditions.

[0120] When logical state quantity When the value is false (i.e., 0), it indicates that the synchronous jigging variance is... Rheological energy ratio with normalized rheological energy They did not reach their respective thresholds simultaneously.

[0121] This means that the disturbances currently affecting the system are generally within the range of fluctuations that the process can cover, or that a single equipment malfunction has not yet evolved into a global material deterioration.

[0122] Under this condition, the state determination module outputs a normal adjustment signal, which maintains the existing normal operating mode of the system and drives the subsequent steady-state adjustment module to perform continuous closed-loop compensation.

[0123] When logical state quantity True (i.e., value 1) and the duration exceeds the set anti-shake time window. When, it indicates the variance of synchronous jigging. Greater than or equal to the jigging disorder variance threshold And the normalized rheological energy ratio Simultaneously greater than or equal to the rheological energy ratio threshold .

[0124] This data indicates that there is a large influx of difficult-to-select materials upstream, while the downstream suspension is already in a complex working condition of high viscosity contamination.

[0125] Faced with such complex operating conditions, conventional control closed loops are prone to integral saturation and regulation failure. Therefore, the control system disconnects the conventional loop and immediately outputs an intervention trigger signal. This intervention trigger signal directly activates the subsequent intervention interlock module, performing forced protection and physical cleaning actions on the entire washing and screening system to mitigate the deterioration of process conditions.

[0126] Regarding the specific electrical action response of the underlying hardware actuator after receiving the intervention trigger signal, those skilled in the art can perform conventional relay logic programming according to the hardware manual of the programmable logic controller, which is a well-known technology in the field and will not be elaborated here.

[0127] See attached document Figure 5 , Figure 5 This is a schematic diagram of the adaptive control and protection execution principle according to an embodiment of the present invention. After receiving the routing control command issued by the state determination module, the steady-state adjustment module and the intervention interlock module provided by the present invention are responsible for converting the abstract logical state into specific underlying device drive signals, thereby realizing the adaptive smooth operation of the washing system and forced protection under extreme operating conditions.

[0128] In this embodiment, the control system specifically executes the following sub-steps to complete the final hardware closed-loop control:

[0129] S401 responds to conventional adjustment signals and performs variable parameter steady-state control;

[0130] When the logic state variable points to the normal adjustment condition, the steady-state adjustment module takes over the system control and maintains the original process closed loop.

[0131] Traditional fixed-parameter proportional-integral-derivative controllers often suffer from overshoot or lag when dealing with heavy suspensions with dynamically fluctuating coal slime content due to changes in fluid resistance.

[0132] To improve the robustness of the control system, the steady-state regulation module uses the normalized rheological energy ratio calculated in the previous step. As a feedforward adaptive adjustment factor, the basic proportional gain of the controller is dynamically corrected. The adaptive proportional gain adjustment formula executed by the control system is as follows:

[0133] ;

[0134] In the formula, This represents the real-time proportional gain after rheological state correction. This represents the initial proportional gain of the control system tuned under clean water calibration conditions. This represents the preset gain attenuation coefficient. Represents the normalized rheological energy ratio; This represents the rheological energy ratio threshold.

[0135] As a preferred method, the gain attenuation coefficient The value is usually set between 0.1 and 0.3 to ensure a smooth transition in gain adjustment.

[0136] To avoid the real-time proportional gain being affected by abnormal system data acquisition or brief shocks from extremely high viscosity. To address the algorithm's dead zone when the calculated value is negative or close to zero, the control system incorporates gain limiting logic. When the calculated value... Below the set safety threshold (e.g.) When the gain is 50% or less, the system forcibly locks the proportional gain of the output for that cycle to a safe lower limit.

[0137] To prevent integral accumulation runaway under this limiting state, the control system activates an anti-integral saturation mechanism simultaneously with the gain limiting, i.e., suspending the accumulation calculation of the integral term within the controller. This collaborative design helps maintain the basic correction capability of the underlying controller in the early stages of parameter deterioration and avoids large oscillations in subsequent adjustments.

[0138] When the system's operating parameters cross the preset safety boundary, a single gain adjustment is no longer sufficient to address the physical risks, and the control strategy must be switched to intervention mode.

[0139] S402, responds to the intervention trigger signal and executes a time-locked physical cleaning action;

[0140] When the status determination module issues an intervention trigger signal, it means that the system is in a complex and severe working condition with the superposition of high viscosity and difficult-to-select materials. At this time, the intervention interlock module will forcibly cut off the closed loop of the steady-state regulation module and take over the underlying equipment according to the preset safety sequence.

[0141] The specific physical actions involve multiple parallel and delayed control branches. The system immediately issues a full-open command to the diversion valve to accelerate the discharge of inferior heavy media in the pipeline network, and simultaneously commands the clean water supply valve to open to its maximum stroke to inject clean water into the pump pool to dilute the suspension.

[0142] Considering the potential for valve jamming in industrial settings, the control system simultaneously reads the physical position feedback signal of the corresponding valve after issuing the aforementioned command.

[0143] If the target opening signal is not detected within the set time limit, the system will trigger a hardware fault interruption alarm and shut down the upstream main feed gate to avoid the spillover risk caused by control logic deadlock.

[0144] During the execution of intervention and protection actions, the control system synchronously monitors the operating status and performs recovery assessments to determine the safe time to exit the protection mode.

[0145] S403, Operating condition reset and mode handover based on hysteresis interval;

[0146] If the system immediately exits the protection state when the parameters are just below the trigger threshold, the operating conditions are often easily deteriorated again due to the inertia of the materials, which can cause high-frequency oscillations in the control mode.

[0147] To mitigate this issue, a hysteresis reset logic was introduced into the intervention interlock module. The control system sets a synchronous jigging variance recovery threshold and a rheological energy ratio recovery threshold, which are typically set as the corresponding trigger thresholds. and 70% to 80%. Only when the synchronous jigging variance... Rheological energy ratio with normalized rheological energy When all values ​​fall below their respective recovery thresholds, and this satisfactory state lasts for more than the set reset debounce window (e.g., 30 to 60 seconds), the intervention interlock module determines that the cleaning and dilution operation is complete. Subsequently, the system cancels the intervention trigger command and smoothly transfers control of the underlying hardware back to the steady-state adjustment module.

[0148] In a step-like transition, the control system executes a smooth switching logic within the same scan cycle during the handover of control.

[0149] Specifically, the system uses the current actual physical output feedback value of the frequency converter and valve at the end of the intervention mode to assign the PID integral term of the steady-state regulation module as the initial starting point, thereby ensuring a smooth transition of control commands and restoring normal production closed-loop regulation.

[0150] See attached document Figure 6 , Figure 6 This is a schematic diagram illustrating the principle of device health assessment and dynamic threshold optimization according to an embodiment of the present invention.

[0151] During the long-term operation of the washing and screening system, mechanical wear of the centrifugal pump impeller and scaling on the inner wall of the pipeline can cause a slow drift in the equipment's reference dynamic response.

[0152] From the perspective of fluid mechanics principles, an increase in the impeller end face clearance often leads to a decrease in the volumetric efficiency inside the pump body. In this case, it requires more shaft power to transport the same flow rate of fluid.

[0153] If the state determination module relies on the initially calibrated fixed threshold for a long time, it is easy to cause misjudgment of the control logic or a decrease in sensitivity.

[0154] To this end, the parameter self-learning and threshold dynamic optimization module provided by the present invention receives historical operation logs and real-time features, and corrects the judgment benchmark online in a data-driven manner.

[0155] In this embodiment, the module specifically performs the following sub-steps to complete the closed-loop iteration of system parameters:

[0156] S501 collects long-term operating parameters of the equipment and constructs a multi-dimensional input feature vector;

[0157] To help assess the degree of physical performance degradation of the equipment, the system needs to extract operational characteristics on a macroscopic time scale and control the system to extract three key business data points within a set assessment period. Specific input items include:

[0158] Cumulative operating time of the equipment since the last overhaul Power drift rate under clean water operation conditions, where the active power deviates from the initial factory calibration value. And the cumulative frequency of system-triggered intervention and protection actions during the past assessment period. .

[0159] Before inputting the data into the evaluation model, the control system performs max-min normalization preprocessing logic on the three data items with different physical dimensions, mapping them to the interval [0, 1], thereby constructing a dimension of Input feature vector These three data points comprehensively reflect the changing trends in the mechanical lifespan of the equipment, the decline in hydraulic efficiency, and the system's resistance to disturbances.

[0160] After the preprocessing stage, this standardized data will be fed into a specific network framework as the underlying basis to deduce specific degradation indicators.

[0161] S502, invoke the device health assessment neural network model and perform forward inference;

[0162] In this embodiment, the system uses a backpropagation neural network with one hidden layer as the evaluation model. The network hierarchy of this model is clearly divided into an input layer, a hidden layer, and an output layer.

[0163] The input layer contains three neurons, each receiving a preprocessed input feature vector. The three elements; the hidden layer is set to 5 neurons, and the Siqmoid activation function is used to handle nonlinear mapping;

[0164] The output layer contains one neuron, employs a linear activation function, and the data stream is passed layer by layer through the weight matrix and bias vector. The output result is the device wear attenuation coefficient. .

[0165] The specific physical and operational meaning of this output represents the percentage decrease in the fluid delivery efficiency of the feed pump compared to its brand-new state. To prevent the model from generating unreasonable predictions under extreme data inputs, the control system adds a limiting operation after acquiring the output, adjusting the equipment wear attenuation coefficient. It is strictly constrained to a value range of 0 to 1.

[0166] To ensure the predictive effectiveness of this neural network model in real-world industrial scenarios, it is necessary to perform detailed offline training and parameter optimization.

[0167] S503 performs offline training of the model and updates network parameters;

[0168] The training samples for the model are derived from the historical operational database accumulated during each major equipment overhaul cycle. The definitions and sources of the sample labels are as follows:

[0169] During the maintenance of historical equipment, maintenance personnel measured the wear of the impeller diameter using vernier calipers, and the ratio of the actual wear to the wear at the limit of scrap was used as the corresponding true label value. Because the time span of manually measured labels is large, while the input feature vector is extracted frequently according to calendar cycles, a label interpolation mapping mechanism is introduced in the data preprocessing stage of the control system in order to construct one-to-one aligned supervised learning sample pairs.

[0170] The system extracts the measured wear labels of two adjacent overhaul nodes and assumes that, in the absence of extreme jamming events, the mechanical wear of the impeller accumulates non-linearly with the material throughput.

[0171] Based on the historical daily processing volume weights, the system uses a cubic spline interpolation algorithm to calculate the virtual wear label value corresponding to each evaluation period, thereby completing the training label set. In the training step, the system uses mean squared error as the loss function. The calculation formula is as follows:

[0172] ;

[0173] In the formula, This represents the mean squared error loss function value. This represents the total number of training samples. The model represents the first Forward inference output value of each sample Indicates the first The system calculates the partial derivatives of the loss function with respect to the weights and biases of each layer using the gradient descent algorithm, and updates the network parameters along the negative gradient direction.

[0174] As a preferred approach, the initial learning rate during the offline training phase can be set to 0.01, and the maximum number of iterations can be set to 1000, until the loss function value converges to below the set training error threshold, thereby completing the offline construction and application deployment of the model.

[0175] See attached document Figure 7 , Figure 7 This is a system hardware deployment and communication topology diagram according to an embodiment of the present invention;

[0176] The control logics and evaluation models involved in this invention need to be implemented based on specific industrial hardware architectures to form a complete closed-loop control network.

[0177] In this embodiment, the system maps the aforementioned functional modules to physical communication links and computing nodes, and specifically performs hardware scheduling and data interaction according to the following sub-steps:

[0178] S601 establishes a physical communication link between the underlying hardware and the control unit;

[0179] Various sensors and actuators deployed in industrial settings directly face complex physical environments. To establish stable data acquisition channels, low-level devices such as electromagnetic flowmeters, frequency converters, and buoy sensors are connected to the input / output modules of programmable logic controllers (PLCs) via standard industrial buses or analog cables. Electromagnetic interference in industrial settings often causes random glitches in analog signals.

[0180] To improve the reliability of subsequent computational features, as a preferred approach, the programmable logic controller (PLC) calls its internal digital filtering algorithm to preprocess the original electrical signal before executing the logic of the data acquisition module.

[0181] From the perspective of signal processing principles, such industrial disturbances are mostly high-frequency spike signals. The control system specifically executes a first-order hysteresis digital filter formula to simulate the smoothing effect of an RC low-pass filter in the hardware circuit.

[0182] ;

[0183] In the formula, This represents the digital filter output value for the current sampling period. This represents the raw physical quantity measurement value transmitted by the underlying sensor in the current sampling period. This represents the digital filter output value of the previous sampling period. This indicates the set filter coefficients.

[0184] The filter coefficient The value is typically set between 0.1 and 0.3, with the specific value fine-tuned based on the frequency of interference at the site. The smoothed signal after digital filtering... This serves as valid basic data, which can be used by the aforementioned data acquisition module and rheological state decoupling extraction module.

[0185] While ensuring the quality of the underlying input data, the control system needs to allocate computing resources reasonably to guarantee the real-time response capability of core actions.

[0186] S602 performs real-time control task allocation at the edge control node;

[0187] The aforementioned state determination module, steady-state adjustment module, and intervention interlock module have high requirements for real-time operation.

[0188] Therefore, the system deploys this part of the control logic, which requires millisecond to second-level response, inside the programmable logic controller in the field.

[0189] The central processing unit (CPU) of the control system executes logic code cyclically according to a set scan cycle. To help prevent CPU overload or watchdog timeout failures caused by multiple concurrent tasks, the control system adopts a time-sharing task scheduling mechanism.

[0190] Specifically, the system configures the adaptive gain calculation of the steady-state adjustment module as a fast task that is executed synchronously with the control cycle;

[0191] Meanwhile, the state determination module, which includes matrix comparison and anti-jitter timing, is configured as a timed interrupt task. This task isolation and prioritization at the physical execution level helps maintain the responsiveness of core protection actions and reduces the risk of conventional calculations blocking critical intervention interlock logic.

[0192] For the multi-task time slice allocation and interrupt priority setting of programmable logic controllers, those skilled in the art can consult the programming manual of the corresponding brand controller. Its task scheduling configuration is a well-known technology in the field and will not be described in detail here.

[0193] Above the basic control level, the system also needs to rely on external computing power to iterate and optimize the algorithm model, thereby constructing a complete distributed collaborative network.

[0194] S603, constructing a collaborative computing architecture between the control node and the upper-level server;

[0195] The parameter self-learning and threshold dynamic optimization module mentioned earlier involves large-scale matrix operations and historical data aggregation, and its computational load usually exceeds the hardware capacity limit of conventional programmable logic controllers. In this embodiment, the system adopts a distributed control system architecture.

[0196] The programmable logic controller on site acts as an edge node, responsible for aggregating and temporarily storing real-time operating characteristics and trigger frequency data;

[0197] The host computer server acts as a computing node, deploying the neural network model and corresponding offline training environment for performing health assessments.

[0198] The edge node and the computing node establish a communication heartbeat via industrial Ethernet. Each time the computing node completes a round of forward inference and threshold optimization calculations according to the set evaluation cycle, it sends the updated rheological energy ratio threshold to a specific register address of the edge node.

[0199] To prevent data errors caused by network fluctuations, edge nodes synchronously execute a data parity check mechanism when they receive a new threshold.

[0200] The edge node only overrides the original judgment criteria when the verification passes and the new threshold is within the preset safety limit. Furthermore, considering the potential for network outages in industrial settings, the control system incorporates a communication degradation protection mechanism within the edge node.

[0201] If the edge node does not receive a heartbeat response from the host computer within the set tolerance window, the system will trigger a communication loss alarm and automatically reject external threshold write commands, forcing the current judgment benchmark to be locked as the valid threshold of the last successful verification.

[0202] Through this design of hardware and software decoupling and communication fault degradation, the system retains the physical security baseline of the underlying control while gaining the ability to evolve strategies with the support of external computing power, thus providing basic operating environment support for the technical features of each layer from physical perception and logical judgment to model optimization.

[0203] To further aid in understanding the technical solution of this invention, the following section provides a complete description of the system's dynamic response and long-cycle optimization process in the context of a specific industrial coal slime washing application scenario, along with experimental verification and effect comparison data.

[0204] At a 10-million-ton-level unpressurized three-product heavy medium cyclone washing site, the control system completed the calibration of benchmark parameters during the initial commissioning phase. In this embodiment, the jigging disorder variance threshold is... The initial rheological energy ratio threshold is set to 0.45. Set to 2.2, under normal operating conditions, the quality of the feed coal is relatively stable, and the synchronous jigging variance is... Rheological energy ratio with normalized rheological energy All values ​​fluctuate at a low level. At this time, the system maintains a steady-state adjustment mode. The feedforward adaptive adjustment factor finely adjusts the proportional gain of the underlying controller in real time, and the suspension density remains near the target set value.

[0205] As the washing and beneficiation process progresses, a sudden change in the upstream coal quality triggers a large influx of high-viscosity fine mud into the system. Bottom-level sensors detect a sharp increase in fluid rheological resistance and a rise in the normalized rheological energy ratio. It climbed to 2.35 within two minutes, with simultaneous elimination variance. It jumped to 0.52.

[0206] The two-dimensional matrix cross-validation results of the state determination module show that the logic state variables From false to true.

[0207] The system did not immediately shut down, but instead started timing the anti-shake time window. After confirmation over four consecutive sampling cycles, the out-of-limit state persisted, thus eliminating physical interference such as transient bubbles.

[0208] Immediately, the intervention interlock module forcibly takes over control, issuing a full-open command to the diversion valve and the clean water supply valve, and simultaneously monitoring the valve feedback opening to prevent jamming.

[0209] After confirming that the action was in place, the system gradually reduced the frequency of the feed pump inverter to slow down the continuous influx of hazardous materials. After about 5 minutes of physical cleaning and dilution, the system parameters returned to below the recovery threshold and smoothly passed the hysteresis anti-shake window. The control system then smoothly returned the current physical output state to the steady-state adjustment module through a non-disruptive switching mechanism, using the current physical output state as the initial integral value. The entire abnormal handling process did not trigger water hammer in the pipeline network.

[0210] After three months of continuous operation, the feed pump impeller exhibited irreversible mechanical wear. The parameter self-learning and threshold dynamic optimization module detected a significant increase in active power drift rate in the background. By invoking a pre-trained health assessment neural network model for forward inference, the system calculated the current equipment wear attenuation coefficient. The value is 0.12. Based on the aforementioned update rules, the system automatically lowers the rheological energy ratio threshold from 2.2 to 2.13. This dynamic adjustment actively tightens the system's tolerance boundary for high-viscosity fluids and compensates for the insufficient delivery capacity caused by the decrease in pump hydraulic efficiency.

[0211] To objectively verify the actual technical effect of the present invention, this embodiment further provides experimental verification and data comparison analysis, as shown in the appendix. Figure 8 , Figure 8 This figure shows a simulation curve comparing the effects of control strategies according to an embodiment of the present invention. The figure is generated based on historical field operation data as input excitation, and the calculated step response and disturbance rejection time series curves of the system under different control strategies are presented.

[0212] In the comparative experiment, the control group used traditional fixed-parameter PID control and a single static fixed threshold protection strategy, while the experimental group used the adaptive control and threshold dynamic optimization strategy described in this invention. After simulating the injection of the same high-viscosity coal slime disturbance signal... Figure 8 The simulation data intuitively reflects the differences between the two. When the control group faced a sudden change in rheological resistance, the controller output an excessive adjustment signal because the proportional gain could not be dynamically attenuated, resulting in an overshoot of 0.08 q / cm³ in the peak density deviation of the suspension. Moreover, after triggering the fixed threshold protection, it took about 14 minutes for the system to recover to a stable state. In contrast, the experimental group's steady-state adjustment module performed adaptive gain correction in advance through the rheological energy ratio, effectively suppressing the peak density deviation to within 0.03 q / cm³. After triggering the intervention protection, relying on the non-disruptive switching mechanism, the system smoothly recovered to the set target density range in only about 6 minutes.

[0213] Furthermore, six months of on-site industrial operation statistics further validated the reliability of this solution. Under traditional control strategies, due to the inability of the judgment benchmark to adapt to the long-term wear of the impeller, four underflow port blockage and pump cavitation events occurred in the middle and later stages of the equipment's life cycle, resulting in a high frequency of unplanned shutdowns. With the solution of this invention, relying on the dynamic adjustment of the rheological threshold using a neural network model, the system proactively eliminated potential critical blockage hazards in the early stages of equipment performance degradation. The incidence of similar serious failures was zero throughout the entire six-month observation period. The above experimental verification and long-term statistical data demonstrate that the control method based on multi-dimensional state characteristics and model self-learning provided by this invention possesses practical engineering application value and significant technical advantages in dealing with complex coal quality disturbances, maintaining process stability, and reducing equipment failure rates.

[0214] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A pressureless three-product heavy medium cyclone reselection system for jigging middlings coal, characterized in that, include: The data acquisition module is used to synchronously acquire the instantaneous physical quantities of the jig and feed pump equipment layers, and perform time dimension translation compensation based on the inherent physical conveying time of the material to obtain the synchronous jig variance after time alignment. The rheological state decoupling extraction module is used to perform reference frequency normalization compensation on the actual electrical parameters based on the instantaneous physical quantities of the feed pump and the law of fluid mechanical similarity, and extract the normalized rheological energy ratio after eliminating frequency conversion interference. The state determination module is used to compare the synchronous jigging variance with the preset jigging disorder variance threshold and the normalized rheological energy ratio with the preset rheological energy ratio threshold. Based on the cross-comparison results of the two-dimensional state matrix, it outputs the operating condition routing control command. The steady-state control module is used to perform variable-parameter steady-state control by using the normalized rheological energy ratio as a feedforward factor when the operating condition routing control command is a conventional control signal. The intervention interlock module is used to cut off the normal closed-loop circuit and perform a time-locked physical cleaning and protection action when the operating condition routing control command is an intervention trigger signal.

2. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 1, characterized in that, The data acquisition module is specifically used to obtain the time-aligned synchronous jigging variance in the following ways: Acquire the instantaneous displacement signal of the upstream jig buoy sensor and calculate the real-time jig variance within a set sliding time window; The ratio of the pre-calibrated physical material transport time to the set high-frequency sampling period is used to calculate the queue depth of the first-in-first-out queue. The real-time jigging variance is continuously stored in the input of the first-in-first-out queue, and the jigging variance after periodic delay is extracted from the output of the first-in-first-out queue and used as the synchronous jigging variance after time alignment.

3. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 1, characterized in that, The rheological state decoupling extraction module is specifically used for extracting the normalized rheological energy ratio after eliminating frequency conversion interference when: The instantaneous active power and instantaneous actual output frequency of the feed pump frequency converter, as well as the instantaneous volumetric flow rate of the electromagnetic flow meter in the feed pipeline, are obtained. The rated power frequency of the feed pump drive motor is used as the reference frequency. The ratio of instantaneous active power to instantaneous volumetric flow rate is calculated. This ratio is then multiplied by the square of the ratio of the reference frequency to the instantaneous actual output frequency to obtain the normalized rheological energy ratio.

4. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 3, characterized in that, The rheological state decoupling extraction module also has a built-in start / stop error prevention mechanism, specifically used for: When the instantaneous volumetric flow rate or the instantaneous actual output frequency is lower than the set start / stop lower limit threshold, the current normalized rheological energy ratio is automatically locked to the calculated value of the previous effective cycle.

5. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 1, characterized in that, The state determination module is internally configured with a stabilization time window for filtering out high-frequency transient interference. When the synchronous jigging variance is less than the jigging disorder variance threshold, or the normalized rheological energy ratio is less than the rheological energy ratio threshold, the state determination module outputs a normal adjustment signal. When the synchronous jigging variance is greater than or equal to the jigging disorder variance threshold, the normalized rheological energy ratio is greater than or equal to the rheological energy ratio threshold, and the duration of exceeding the limit exceeds the anti-jigging time window, the state determination module determines that the system is in a compound severe working condition and outputs an intervention trigger signal.

6. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 1, characterized in that, The steady-state adjustment module is specifically used for: when performing variable parameter steady-state control. The attenuation term is constructed by using the ratio of the normalized rheological energy ratio to the rheological energy ratio threshold, and the basic proportional gain is dynamically corrected to obtain the real-time proportional gain. When the real-time proportional gain is lower than the set safety lower limit, the proportional gain of the current cycle is forcibly locked to the safety lower limit, and the accumulation calculation of the internal controller integral term is simultaneously suspended.

7. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 1, characterized in that, The intervention interlock module is specifically used for: When executing the time-interlocked physical cleaning protection action, it is used to: The system sends opening commands to the diversion valve and clean water supply valve of the washing pipeline and simultaneously reads the physical position feedback signals of the corresponding valves. If the target opening signal is not detected within the set time limit, a hardware fault interruption alarm will be triggered, and the upstream main feed gate will be shut down in conjunction with the alarm.

8. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 7, characterized in that, The intervention interlock module is specifically used to: When restoring system control, it is used to: When the synchronous jigging variance and the normalized rheological energy ratio both fall below the set recovery threshold, and the duration of this satisfactory state exceeds the set reset anti-shake window, the intervention trigger signal is cancelled. The actual physical output feedback value of the underlying actuator is synchronously assigned to the integral term of the steady-state regulation module as the initial starting point, and the system control is transferred.

9. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 1, characterized in that, The system also includes a pre-filtering module and a task scheduling unit; The pre-filtering module is used to call a first-order hysteresis digital filtering algorithm to perform smooth preprocessing on the original electrical signal. The task scheduling unit is configured within the programmable logic controller to configure the steady-state adjustment module to run as a fast task that is executed synchronously with the control cycle, and to configure the cross-comparison of the two-dimensional state matrix of the state determination module as a timed interrupt task.

10. The jigging middlings unpressurized three-product heavy medium cyclone reselection system according to claim 1, characterized in that, The system also includes a health assessment and threshold optimization module, used for: Within the set evaluation period, the cumulative running time of the equipment since the last overhaul, the power drift rate under clean water operation conditions, and the cumulative frequency of historical trigger intervention protection actions are extracted, and max-min normalization preprocessing is performed to construct a multi-dimensional input feature vector. The multidimensional input feature vector is fed into the backpropagation neural network to perform forward inference, calculate and limit the output of the equipment wear attenuation coefficient that characterizes the deterioration of the conveying efficiency, and dynamically adjust the current rheological energy ratio threshold accordingly. The method for obtaining the verification label set during offline training of the backpropagation neural network is as follows: The actual impeller wear measured during historical overhauls is used as the true label value. Combined with the weight of the daily material handling volume in history, the virtual wear label value corresponding to each evaluation cycle is calculated using a cubic spline interpolation algorithm.

Images