High-end automatic ash conveying big data energy-saving control method and system

Abstract

The invention discloses a high-end automatic ash conveying big data energy-saving control method and system, and belongs to the technical field of energy-saving ash conveying. The method specifically comprises the following steps; according to the method, the pipeline resistance coefficient and the ash-gas ratio are calculated through visual measurement and multi-source time sequence data fusion, a pressure flow equation is constructed, a bin pump pressure change curve and an air consumption curve for maintaining stable conveying of the ash plug in a future limited time domain are generated, and the optimal ash-gas ratio is used as a guide; establishing a conveying cost function about the air inlet valve position increment and the pressure tracking deviation, solving the conveying cost function to obtain an optimal air inlet valve position sequence in a future finite time domain, and performing energy-saving ash conveying control; on the premise of ensuring the conveying reliability, the compressed air consumption of the unit ash conveying amount and the average load of the ash conveying system are reduced, blind throttling is avoided, and the collaboration of the ash conveying system and other production links is enhanced.

Description

Technical Field

[0001] This invention relates to a high-end automatic ash conveying big data energy-saving control method and system, belonging to the field of energy-saving ash conveying technology. Background Technology

[0002] Pneumatic ash conveying is a process that uses airflow as a conveying medium to transport bulk materials from one or more sources to one or more destinations. This process is a key technology for power plants and chemical plants to handle powdery and granular materials. In thermal power plants, ash conveying pipelines can transport fly ash produced by boiler combustion. In the process of building materials production, ash conveying pipelines can transport building materials such as cement and limestone powder. Therefore, ash conveying capacity is also a direct reflection of the economic benefits of enterprises.

[0003] In existing technologies, the control methods for pneumatic ash conveying typically employ fuzzy control or expert systems. By pre-setting pressure time or material level time rules, the automatic sequential start-up and shutdown of the silo pump and the backflushing frequency adjustment are achieved. However, existing technologies have the following shortcomings: timed or differential pressure backflushing wastes compressed air and affects conveying; there is a lack of coordination with other production processes; and under light load conditions, reducing the pumping volume by closing the inlet valve or bypassing the return flow results in wasted energy due to throttling and heat generation. Summary of the Invention

[0004] The purpose of this invention is to provide a high-end automatic ash conveying big data energy-saving control method and system to solve the problems of high ash conveying cost, instability and lack of coordination in the existing technology.

[0005] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0006] High-end automatic ash conveying big data energy-saving control methods include: Upstream production data, ash conveying process data, and downstream power data of the ash conveying pipeline network are collected by visual inspection. The pipeline resistance coefficient and ash-to-gas ratio are calculated by soft measurement estimation. The upstream production data includes ash hopper level change data. By inputting the pipeline resistance coefficient and the ash-to-gas ratio into the pressure-flow equation, the pressure change curve of the silo pump and the gas consumption curve for maintaining the ash plug delivery within a future finite time domain are generated. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure and inlet flow. Based on the pressure change curve and gas consumption curve of the silo pump, a transport cost function for the increment of the inlet valve position and the pressure tracking deviation is established, and the optimal inlet valve position sequence in the future finite time domain is obtained by solving it. The optimal intake valve position sequence is sent to the intake valve for execution and the corresponding silo pump is adjusted. Based on the pressure deviation during the execution process, the optimization command is used to carry out energy-saving control of the ash conveying pipeline network.

[0007] Specifically, upstream production data, ash conveying process data, and downstream power data of the ash conveying pipeline network are collected through visual inspection. Soft sensing estimation is used to calculate the pipeline resistance coefficient and ash-to-gas ratio. The upstream production data includes ash hopper level change data, including: The surface image sequence of the material in the ash hopper is collected by industrial vision sensors from upstream production data. The grayscale and texture features are identified by visual inspection to establish a three-dimensional contour of the material surface, and the material level change data of the ash hopper is measured based on the three-dimensional contour of the material surface. The upstream production data, ash conveying process data, and downstream power data are time-series aligned, and outlier data points are removed to obtain the conveying sample set; Based on the transport sample set, the pressure stability characteristics of the silo pump during the feeding stage, the pressure decay gradient during the transport stage, and the valve position time sequence integral value of the air inlet valve are calculated. Combined with the pipeline resistance coefficient as an unknown, a pressure decay differential equation is constructed. The pipeline resistance coefficient is obtained based on the pipeline geometry and state information. The root mean square error is calculated based on the pressure prediction value obtained by solving the pressure attenuation differential equation and the pressure change parameters in the ash conveying process data. The pipeline resistance coefficient is then adjusted iteratively using gradient descent until the root mean square error is less than the preset error threshold, thus obtaining the optimized pipeline resistance coefficient. The optimized pipeline resistance coefficient is substituted into the soft sensor estimation equation to calculate the ash-to-gas ratio of the current conveying cycle. The soft sensor estimation equation is derived based on the preset material conservation equation and gas state equation.

[0008] Specifically, based on the transport sample set, the pressure stability characteristics during the silo pump feeding stage, the pressure decay gradient during the transport stage, and the valve position time-series integral value of the inlet valve are calculated, including: Based on the pressure time series data of the silo pump during the feeding stage extracted from the transport sample set, its pressure variance was calculated as a pressure stationary feature. Based on the transport sample set, the pressure of the silo pump is identified to enter the monotonically decreasing transport stage. Linear piecewise fitting is performed on the decreasing stage, and the resulting slope is used as the pressure decay gradient. The timing data of the valve position command and feedback signal of the intake valve are extracted from the transported sample set, and the corresponding valve position timing integral value is obtained by numerical integration.

[0009] Specifically, the pipeline resistance coefficient and ash-to-gas ratio are input into the pressure-flow equation to generate the pressure change curve and gas consumption curve of the silo pump for maintaining ash plug delivery within a finite time domain. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure, and inlet flow rate, and includes: The conveying sample set is resampled according to a preset time step, taking the conveying cycle of the ash conveying pipeline as the unit, to generate the inlet valve position sequence, the silo pump pressure sequence, and the inlet flow rate sequence. The intake valve position sequence, the silo pump pressure sequence, and the intake flow rate sequence are used as inputs to a preset discrete-time difference equation containing an undetermined coefficient matrix. The pressure-flow equation in time-varying parameter form is determined by recursively updating the undetermined coefficient matrix. The pressure-flow equation in time-varying parameter form is obtained by substituting the pipe resistance coefficient as a correction factor into the pressure-flow equation and establishing a Jacobian matrix for the pipe resistance coefficient to update the parameters in the pressure-flow equation in time-varying parameter form. Using the ash-to-gas ratio as the gain coefficient, a set of differential equations is constructed by simultaneously solving the pressure-flow equations for the pressure change rate of the silo pump and the inlet flow rate change rate. Using the current silo pump pressure, inlet valve position, and the initial solution state of the differential equation system as initial conditions, numerical integration of the differential equation system is performed to generate the silo pump pressure change curve and gas consumption curve for ash hydrant transportation within a finite time domain in the future.

[0010] Specifically, the intake valve position sequence, the hopper pump pressure sequence, and the intake flow rate sequence are used as inputs to a preset discrete-time difference equation containing a matrix of undetermined coefficients. By recursively updating the matrix of undetermined coefficients, the time-varying parameter form of the pressure-flow equation is determined, including: Based on the intake valve position sequence, the chamber pump pressure sequence, and the intake flow rate sequence, a discrete-time difference equation containing an undetermined coefficient matrix is ​​constructed as the basic expression of the pressure-flow rate equation. The predicted output term of the silo pump pressure in the discrete-time difference equation is subtracted from the actual measured value of the silo pump pressure sequence, and the error accumulation function about the undetermined coefficient matrix is ​​established through the obtained residual sequence. The normal equation that makes its gradient zero is derived based on the error accumulation function, and the element values ​​of the undetermined coefficient matrix are iteratively updated until the change in the coefficient matrix of adjacent iteration steps is less than the preset convergence threshold, thus obtaining the converged coefficient matrix. Substituting the coefficient matrix after iterative convergence into the discrete-time difference equation, a pressure-flow equation in time-varying parameter form is generated.

[0011] Specifically, the normal equation whose gradient is zero is derived based on the error accumulation function, and the element values ​​of the undetermined coefficient matrix are iteratively updated until the change in the coefficient matrix in adjacent iterations is less than a preset convergence threshold, resulting in the converged coefficient matrix, including: By taking the partial derivatives of the elements of the undetermined coefficient matrix through the error accumulation function, we obtain the gradient vector expression composed of the first-order partial derivatives. By setting the gradient vector expression equal to the zero vector, we establish a matrix equation with the undetermined coefficient matrix as the unknown quantity. The matrix equation is decomposed into undetermined coefficient matrix and vector terms, and incrementally updated by a preset attenuation factor to obtain the normal equation. The vector terms are composed of the actual measured values ​​of the intake valve position sequence, intake flow sequence and silo pump pressure sequence. Substitute the undetermined coefficient matrix into the normal equation as the initial solution to calculate the residual vector and the Jacobian matrix. A system of linear equations is established using the Jacobian matrix as coefficients and the residual vector as the right-hand side. The correction amount of the undetermined coefficient matrix is ​​obtained by solving the system of linear equations, and the undetermined coefficient matrix is ​​iteratively updated. The L2 norm of the difference between the corresponding elements of the undetermined coefficient matrix of the current iteration step and the previous iteration step is compared with the preset convergence threshold. If it is greater than the convergence threshold, the iteration update continues. If it is less than or equal to the convergence threshold, the current undetermined coefficient matrix is ​​used as the converged coefficient matrix.

[0012] Specifically, based on the pressure change curve and air consumption curve of the silo pump, a delivery cost function is established regarding the increment of the intake valve position and the pressure tracking deviation. Solving this function yields the optimal intake valve position sequence within a future finite time domain, including: Based on the pressure change curve and gas consumption curve of the silo pump, determine the pressure reference value and allowable gas consumption threshold at different future times, and calculate the minimum safe pressure curve in the future limited time domain by combining the pipeline resistance coefficient, ash-to-gas ratio and ash hopper level change data. Based on the minimum safe pressure curve and the theoretical minimum gas consumption, a pressure reference curve and an allowable gas consumption threshold curve are obtained by compensating for the preset process safety margin. The theoretical minimum gas consumption is calculated based on the ash-to-gas ratio and the ash conveying rate. The pressure reference curve and the pressure change curve of the silo pump are subtracted point by point to obtain the pressure tracking deviation sequence. The intake valve position increment sequence is initialized as the decision vector. The delivery cost function is constructed based on the pressure tracking deviation sequence and the decision vector. The minimum safe pressure curve and the total gas consumption upper limit obtained by integrating the allowable gas consumption threshold curve are used as boundary conditions, which are transformed into a set of linear inequalities to form a set of linear constraints in the future finite time domain. The influence of the decision vector on the transport cost function value is calculated within the feasible region defined by the linear constraint set. The value of the decision vector that satisfies the linear constraint set and minimizes the transport cost function value is taken as the optimal intake valve position sequence in the future finite time domain.

[0013] Specifically, within the feasible region defined by the linear constraint set, the impact of the decision vector on the transport cost function value is calculated. The decision vector that satisfies the linear constraint set and minimizes the transport cost function value is taken as the optimal intake valve position sequence in the future finite time domain, including: The feasible region of the decision vector is determined based on the set of linear constraints, and the current intake valve position is used as the initial point for the iterative search. Within the feasible region, the gradient information of the decision vector is obtained by taking the partial derivative of the cost function with respect to the decision vector at the current iteration point. Based on this, a direction vector pointing in the descent direction and within the feasible region is constructed. The decision vector is updated along the direction vector at the initial point according to the preset iteration step size, and the pressure tracking deviation sequence corresponding to the updated decision vector is calculated. Substitute the updated pressure tracking deviation sequence into the transport cost function, calculate the transport cost function value, and iteratively update the position of the decision vector in the feasible region. For the decision vector that minimizes the transport cost function and satisfies all linear constraints, it is converted into the optimal intake valve position sequence in the future finite time domain.

[0014] Specifically, the optimal intake valve position sequence is sent to the intake valve for execution and the corresponding silo pump is adjusted. Based on the pressure deviation optimization command during the execution process, energy-saving control is performed on the ash conveying pipeline network, including: Using the minimum pressure required to maintain stable delivery of ash plugs as a benchmark, the intake valve position values ​​in the optimal intake valve position sequence are converted into energy-saving control commands to drive the intake valves. The gas consumption curve and the pressure change curve of the silo pump within a limited time domain in the future are sent as feedforward values ​​to the silo pump to adjust its output power. The actual pressure inside the current silo pump is obtained, and the instantaneous pressure deviation signal reflecting the movement state of the ash plug and the pipeline resistance is generated by combining the energy-saving control command execution result of the air intake valve. The current conveying efficiency and air consumption status of the ash hydrant are determined based on the instantaneous pressure deviation signal, and the air inlet valve position is adjusted to achieve energy-saving ash conveying control.

[0015] The high-end automatic ash conveying big data energy-saving control system includes a working condition sensing module, an energy efficiency measurement module, a conveying optimization module, and an execution module. The working condition perception module is used to collect upstream production data, ash conveying process data and downstream power data of the ash conveying pipeline network through visual detection, and calculate the pipeline resistance coefficient and ash-to-gas ratio through soft measurement estimation. The upstream production data includes ash hopper level change data. The energy efficiency measurement module is used to input the pipeline resistance coefficient and ash-to-gas ratio into the pressure-flow equation to generate the pressure change curve and gas consumption curve of the silo pump to maintain ash plug delivery within a future finite time domain. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure and inlet flow. The delivery optimization module is used to establish a delivery cost function for the inlet valve position increment and pressure tracking deviation based on the pressure change curve and gas consumption curve of the silo pump, and solve it to obtain the optimal inlet valve position sequence in the future finite time domain. The execution module is used to send the optimal intake valve position sequence to the intake valve for execution and adjust the corresponding silo pump. Based on the pressure deviation optimization command during the execution process, it performs energy-saving control on the ash conveying pipeline network.

[0016] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: By integrating and analyzing data from the entire ash conveying process, the pressure change curve and air consumption curve of the silo pump during ash plug conveying within a finite time domain in the future are predicted. A conveying cost function for the incremental air inlet valve position and pressure tracking deviation is constructed. By solving the optimal air inlet valve position sequence within the finite time domain in the future, energy-saving control of the ash conveying pipeline network is achieved. This invention solves the problems of high ash conveying cost, insufficient stability and coordination in existing technologies. Under the premise of ensuring conveying reliability, it reduces the compressed air consumption per unit ash conveying volume and the average load of the ash conveying system, avoids blind throttling, and enhances the coordination between the ash conveying system and other production links.

[0017] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of this disclosure. Attached Figure Description

[0018] Figure 1 Flowchart of the high-end automatic ash conveying big data energy-saving control method provided by the present invention; Figure 2 A schematic diagram of the ash conveying cycle process provided by the present invention; Figure 3 This is a schematic diagram of material movement provided by the present invention; Figure 4 The structural diagram of the high-end automatic ash conveying big data energy-saving control system provided by the present invention. Detailed Implementation

[0019] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0020] The term "and / or" simply describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0021] Example 1: Please see Figures 1-3 This invention provides an embodiment of a high-end automatic ash conveying big data energy-saving control method, which includes the following specific steps: Step S1: Collect upstream production data, ash conveying process data, and downstream power data of the ash conveying pipeline network through visual inspection. Calculate the pipeline resistance coefficient and ash-to-gas ratio through soft measurement estimation. The upstream production data includes ash hopper level change data.

[0022] The specific steps of step S1 are as follows: Step S101: Collect surface image sequences of materials in the ash hopper from upstream production data using industrial vision sensors, establish a three-dimensional profile of the material surface by visual inspection and identification of its grayscale and texture features, and measure the ash hopper material level change data of the three-dimensional profile of the material surface. In this embodiment, an industrial vision sensor acquires surface images of the material inside the ash hopper at a fixed frame rate, e.g., 1 frame per second, generating a continuous grayscale image sequence. Each image is, for example, 640 pixels wide, 480 pixels high, and has one grayscale channel. The fly ash accumulation surface in the image appears as an area with specific light and dark distribution and granular texture, while the background of the ash hopper inner wall has uniform texture features. A U-Net semantic segmentation model based on an encoder and decoder structure is used to identify the grayscale and texture features of the image. The input to the U-Net semantic segmentation model is a single-frame image. The encoder part consists of four downsampling stages, each containing two convolutional layers and one max-pooling layer, using the ReLU activation function to progressively extract multi-scale features of the image. Shallow features include edges and point textures, while deep features include the overall accumulation shape and the global contrast relationship between the material area and the background area. The decoder part consists of four upsampling stages, each upsampling through transposed convolution and then fused with the high-resolution feature map of the corresponding stage of the encoder through skip connections to restore spatial details and accurately locate boundaries. The final output layer uses a single... Convolution combined with the Sigmoid activation function generates a binary segmentation mask of the same size as the input image, with a size of [missing value]. In a binary segmentation mask, a pixel value of 1 represents the fly ash material area, and a pixel value of 0 represents the ash hopper wall area. The binary segmentation mask of each frame is combined with a known observation geometry model. Using triangulation, the pixels on the material surface contour line in the binary segmentation mask are transformed from two-dimensional image coordinates to a three-dimensional hopper coordinate system, reconstructing the three-dimensional spatial contour point cloud of the material surface. The average coordinates of the highest and lowest points in the material level height direction are calculated based on this three-dimensional spatial contour point cloud, which is used as the current ash hopper material level height. This process is repeated by continuously recording the ash hopper material level height and calculating its value in the direction of the material level height. The rate of change of material level is obtained by measuring the change per unit time, such as per minute. The material level height in the hopper is combined with the rate of change of material level to obtain the material level change data in the hopper. It should be noted that the data for constructing the U-Net semantic segmentation model covers multiple states such as empty hopper, half hopper, full hopper, and material surface collapse during unloading, making it robust to dynamically changing material surfaces. The pixel intensity and texture pattern in the image are mapped to the probability of material presence, thereby accurately separating the material surface contour. The observation geometry model is the fixed installation position and angle of the industrial vision sensor relative to the hopper wall.

[0023] Step S102: Perform time-series alignment on upstream production data, ash conveying process data, and downstream power data, and remove outlier data points to obtain the transport sample set.

[0024] In this embodiment, upstream production data is collected using level gauges, ash conveying process data is collected using pressure transmitters and mass flow meters, and downstream power data is collected using pressure sensors and flow meters. An interpolation-based method is used to align the upstream production data, ash conveying process data, and downstream power data to a unified time grid, for example, one data point every 100 milliseconds. Outlier data points are then removed from the time-aligned data based on statistical distribution. For example, for the pressure parameters of the silo pump, the moving average and standard deviation of the data are calculated within a sliding time window, and outlier data points deviating from the average by more than three times the standard deviation and not conforming to the physical laws of the process are removed. Data points marked as outliers are removed, and spline interpolation is used to fill in the gaps based on the normal data before and after them, forming a continuous and smooth transport sample set. The quality of the transport sample set determines the reliability and stability of the energy-saving control effect. Upstream production data includes the material level change curve of the ash hopper and ash characteristics. Ash transport process data includes the inner diameter and length of the pipeline, the valve position data of the air inlet valve, the number of elbows, the high-frequency pressure signal in the silo pump, the valve position feedback signal of the air inlet valve, and the reading of the air inlet flow meter. Downstream power data includes the pressure and flow of the air compressor main pipe, and process physical laws such as the step change of pressure during non-operation periods.

[0025] exist Figure 2In the process, the conveying process includes the upstream production stage, the ash conveying stage, and the downstream power stage. The ash conveying stage is specifically implemented by the ash conveying pipeline, which uses the pressure energy of compressed air as energy and converts the pressure energy of compressed air into the mechanical energy required for production. Several air inlet valves are distributed on the ash conveying pipeline. The air inlet valves are used to adjust the actual ash conveying action and control the pressure, flow rate, and flow direction of the compressed air.

[0026] Step S103: Calculate the pressure stability characteristics of the silo pump during the feeding stage, the pressure decay gradient during the conveying stage, and the valve position time-series integral value of the air inlet valve based on the conveying sample set. Combine the pipeline resistance coefficient as an unknown quantity to construct a pressure decay differential equation. The pipeline resistance coefficient is obtained based on the pipeline geometry and state information.

[0027] In this embodiment, the start and end points of a complete conveying cycle are identified. The opening of the silo pump feed valve is taken as the start of the conveying cycle, and the end of purging and pressure reset is taken as the end of the conveying cycle. The pressure stability characteristics of the silo pump feed stage, the pressure decay gradient of the conveying stage, and the valve position time-series integral value of the air inlet valve are calculated. Time is taken as the independent variable, and the silo pump pressure is taken as the dependent variable. The pressure decay gradient, instantaneous pressure, air intake conditions related to the valve position time-series integral value, and pipeline resistance coefficient are combined to construct a pressure decay differential equation. It should be noted that the pressure decay differential equation is based on fluid mechanics and the principle of gas-solid two-phase flow. Its form expresses the process of driving force being consumed to overcome pipeline resistance and accelerate ash plugs. The pipeline resistance coefficient is calculated based on the pipeline inner diameter, length, air inlet valve position data, ash characteristics, and number of bends. The pipeline resistance coefficient is a time-varying parameter used to represent the friction of the pipeline inner wall, local resistance, and the current physical characteristics of the ash plugs.

[0028] Step S103 calculates the pressure stability characteristics during the silo pump feeding stage, the pressure attenuation gradient during the conveying stage, and the valve position time-series integral value of the inlet valve based on the conveying sample set, including: Based on the pressure time series data of the silo pump during the feeding stage extracted from the transport sample set, its pressure variance is calculated as a pressure stability feature.

[0029] In this embodiment, the closure of the silo pump exhaust valve and the opening of the feed valve are taken as the starting point, and the closure of the feed valve or the high level signal of the level gauge are taken as the ending point. All silo pump pressure measurement values ​​with aligned timestamps within this time period are extracted to form a pressure sequence for the feeding stage. The average value of the pressure sequence for the feeding stage is calculated, and the variance is obtained by averaging the squares of the differences between each pressure value and the average value. A small variance value indicates that the silo pump has good airtightness during the feeding stage, with no abnormal leakage or pressure shock, and also indicates that there is no waste of compressed air due to leakage.

[0030] Based on the transport sample set, the pump pressure is identified as entering a monotonically decreasing transport phase. Linear piecewise fitting is performed on this decreasing phase, and the resulting slope is used as the pressure decay gradient.

[0031] In this embodiment, after the air inlet valve is opened, the conveying stage begins. After the initial pressurization, the pressure of the silo pump enters an approximately linear monotonically decreasing period as the ash plugs are discharged. During the monotonically decreasing period, the inflection point from the pressure peak to the point of stable decrease is located. The period from the inflection point until the pressure drops to near the residual resistance of the pipeline is defined as the stable decreasing stage. The data points of the stable decreasing stage are divided into continuous and overlapping small intervals. Least squares linear fitting is performed on each small interval to obtain a local slope sequence. The median of the local slope sequence is taken to obtain the pressure decay gradient representing the stable conveying process. A negative gradient indicates that the pressure drops rapidly per unit time and the ash plugs move quickly. A gentle gradient indicates that the resistance increases due to ash accumulation in the pipeline, ash stickiness, or excessively high conveying concentration, requiring a longer period of high pressure or a larger flow rate, resulting in increased energy consumption.

[0032] The timing data of the valve position command and feedback signal of the intake valve are extracted from the transported sample set, and the corresponding valve position timing integral value is obtained by numerical integration.

[0033] In this embodiment, valve position data of the intake valve is extracted from the beginning to the end of the complete time period of the conveying stage, including the valve position command signal of the intake valve, the target valve position value issued by the control system, the valve feedback signal, and the actual valve position value returned by the valve positioner. A valve position time series is established, and the area under the valve position time series curve is calculated by numerical integration using the trapezoidal rule. The complete time period is divided into multiple time points, and the valve position values ​​of two adjacent time points are averaged and multiplied by the time interval between the two points to obtain the area of ​​a tiny trapezoid. The areas of all tiny trapezoids are accumulated to obtain the total integral value of the entire time period, which represents the cumulative intensity of the valve's airflow action to complete this conveying. By comparing and analyzing the integral values ​​of different conveying cycles when conveying similar ash amounts, the economic efficiency of valve operation is judged.

[0034] Step S104: Calculate the root mean square error based on the pressure prediction value obtained by solving the pressure attenuation differential equation and the pressure change parameters in the ash conveying process data. Use gradient descent to iteratively adjust the pipeline resistance coefficient until the root mean square error is less than the preset error threshold, and obtain the optimized pipeline resistance coefficient.

[0035] In this embodiment, the measured pressure value of the silo pump, the valve position data of the intake valve, and the pipeline resistance coefficient are substituted into the pressure decay differential equation to calculate the rate of change of pressure with time at the current moment. A preset time interval is used, and the current measured pressure value of the silo pump is added to the product of the rate of change and the time interval to obtain the predicted pressure value at the next time point. This calculation process is repeated until the entire conveying period is covered, ultimately generating a silo pump pressure prediction sequence from the beginning to the end of the conveying stage. The square of the difference between each predicted silo pump pressure value and the actual measured silo pump pressure value is calculated. All squared differences are accumulated over the entire conveying period and averaged. The square root of the average value is taken to obtain the root mean square error, which reflects the overall prediction accuracy. The root mean square error quantifies the degree of matching between the currently assumed pipeline resistance coefficient and the actual system. The gradient of the change in the pipeline resistance coefficient with respect to the current root mean square error is calculated. The gradient is used to indicate the direction and magnitude of adjustment of the pipeline resistance coefficient to reduce the error. The pipeline resistance coefficient is adjusted along the gradient descent direction to obtain an estimate. The new root mean square error is calculated iteratively using the estimate until the difference between the root mean square errors calculated in two iterations is less than a preset error threshold. The optimized pipeline resistance coefficient is then obtained. The time interval step size is set by those skilled in the art according to the actual situation, and the error threshold is set according to the ash conveying pressure control accuracy requirements, usually 0.03 MPa.

[0036] In this example, the initial estimate of the pipeline resistance coefficient is set to an empirical value, such as 0.05. The input data includes the actual silo pump pressure measurements extracted from the transport sample set for this transport stage, and the intake valve position sequence representing the intake conditions. Substituting the input data into the pressure decay differential equation, a numerical method is used to progressively calculate from the start of transport, generating a silo pump pressure prediction sequence. The difference between the predicted and actual values ​​at each moment is calculated for the silo pump pressure prediction sequence and the actual silo pump pressure measurements. The squares of all differences are summed, averaged, and the square root is taken to obtain the initial root mean square error, for example, 0.8 MPa. This error indicates... The initial estimated pipe resistance coefficient caused the overall predicted pressure to be too high. The gradient direction of the error with respect to the pipe resistance coefficient was analyzed, and the pipe resistance coefficient was adjusted to 0.04 along this direction by a preset learning step size, such as 0.01. The above prediction and error calculation process was repeated to obtain a smaller error value of 0.3 MPa. The pipe resistance coefficient was adjusted to 0.03 for the third time, and the calculation error was reduced to 0.1 MPa. When the pipe resistance coefficient was adjusted to 0.02 in the fourth iteration, the error obtained was 0.02 MPa, which was less than the preset error threshold of 0.03 MPa. The iteration stopped, and the optimized pipe resistance coefficient of 0.02 was output.

[0037] Step S105: Substitute the optimized pipeline resistance coefficient into the soft sensor estimation equation to calculate the ash-to-gas ratio of the current conveying cycle. The soft sensor estimation equation is derived based on the preset material conservation equation and gas state equation.

[0038] In this embodiment, the soft sensor estimation equation is derived based on the material conservation equation and the gas state equation. The material conservation equation states that the mass of ash delivered is equal to the mass of ash entering the silo pump. The gas state equation converts the cumulative intake volume measured under actual operating conditions into the gas volume under standard conditions using the average pressure and average temperature of the current conveying process. Based on the quantitative relationship between standard state gas consumption, pipeline conveying resistance, and the mass of ash delivered, the ash-to-gas ratio is used as the unknown quantity in the soft sensor estimation equation. The optimized pipeline resistance coefficient, the cumulative intake flow rate measured in the current conveying cycle, and the average pressure and average temperature of the conveying process are substituted into the soft sensor estimation equation to calculate the ratio of average pressure to average temperature to correct the gas volume. This ratio is then multiplied and divided by the cumulative intake flow rate and the pipeline resistance coefficient to obtain the ash-to-gas ratio of the current conveying cycle. The ash-to-gas ratio represents the ratio of the total mass of ash delivered in this conveying cycle to the total standard volume of compressed air consumed. The higher the value, the more ash is delivered per unit of air, and the higher the energy efficiency.

[0039] Step S2: Input the pipeline resistance coefficient and ash-to-gas ratio into the pressure-flow equation to generate the pressure change curve and gas consumption curve of the silo pump to maintain ash plug delivery within a finite time domain in the future. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure and inlet flow rate.

[0040] The specific steps of step S2 are as follows: Step S201: Resample the conveying sample set according to a preset time step, taking the conveying cycle of the ash conveying pipeline as the unit, to generate the inlet valve position sequence, the silo pump pressure sequence, and the inlet flow rate sequence.

[0041] In this embodiment, the conveying cycle data segment is divided from the conveying sample set according to time markers. Each conveying cycle data segment covers the period from the start of feeding to the end of purging. Resampling is performed on each conveying cycle data segment at preset equal time intervals. Based on their original data points, spline interpolation is used to calculate smoothed estimates for the inlet valve position, silo pump pressure, and inlet flow rate at each equal time interval, resulting in a sequence of inlet valve position, silo pump pressure, and inlet flow rate with the same number of data points arranged at a fixed step size. The equal time intervals are set by those skilled in the art according to actual conditions. The duration of the future finite time domain is determined based on the ash hopper level of the current conveying cycle and the predicted conveying stage duration of the ash-to-air ratio.

[0042] Step S202: The intake valve position sequence, the silo pump pressure sequence, and the intake flow rate sequence are used as inputs to a preset discrete-time difference equation containing an undetermined coefficient matrix. The pressure and flow rate equation in time-varying parameter form is determined by recursively updating the undetermined coefficient matrix.

[0043] The specific steps of step S202 are as follows: Step S2021: Based on the intake valve position sequence, the chamber pump pressure sequence, and the intake flow rate sequence, a discrete-time difference equation containing an undetermined coefficient matrix is ​​constructed as the basic expression of the pressure-flow rate equation.

[0044] In this embodiment, data points from different times arranged in chronological order in the intake valve position sequence, the silo pump pressure sequence, and the intake flow rate sequence are combined to form an augmented historical state vector. A discrete-time difference equation containing a matrix of undetermined coefficients is constructed. The augmented historical state vector is used as input, and the predicted silo pump pressure at the current time is used as output. The discrete-time difference equation is used as the basic expression of the pressure-flow equation. The elements of the matrix of undetermined coefficients are initialized to zero or near-zero random numbers. The dimension of the matrix is ​​determined by the order of the pressure-flow equation. The elements of the matrix of undetermined coefficients represent the influence weight of the corresponding historical state variable on the current pressure prediction.

[0045] Step S2022: Subtract the predicted output term of the silo pump pressure in the discrete-time difference equation from the actual measured value of the silo pump pressure sequence, and establish an error accumulation function about the undetermined coefficient matrix through the obtained residual sequence.

[0046] In this embodiment, the predicted output term of the silo pump pressure in the discrete-time difference equation is compared point by point with the actual measured value of the silo pump pressure sequence. Using the currently assumed undetermined coefficient matrix, combined with the intake valve position sequence and intake flow sequence, the predicted value of the silo pump pressure at each sampling time from the beginning to the end is recursively calculated through the discrete-time difference equation. The actual measured value in the silo pump pressure sequence is subtracted from the predicted value of the silo pump pressure at the corresponding time in the silo pump pressure prediction sequence to establish a residual sequence. An error accumulation function is constructed based on the residual sequence and the exponential decay rule. The error accumulation function is used to evaluate the quality of the undetermined coefficient matrix. The exponential decay rule is set so that the weight coefficient of the residual that is further away from the current calculation time decays smaller according to the exponential law, and the weight of the more recent residual is larger.

[0047] In this example, we assume that data from three consecutive sampling times (t1, t2, t3) are selected in a delivery cycle for illustration. The input data is the intake valve position sequence [40%, 45%, 48%], and the actual measured values ​​of the silo pump pressure sequence are [0.45MPa, 0.48MPa, 0.46MPa]. Assuming that the current undetermined coefficient matrix is ​​initially determined based on historical data, the predicted values ​​of the silo pump pressure at these three times are calculated using the undetermined coefficient matrix, the intake valve position sequence, and the intake flow rate sequence: [0.44MPa, 0.47MPa, 0.49MPa]. The residual is obtained by subtracting the predicted value from the actual measured value at each time point. The sequence is [+0.01MPa, +0.01MPa, -0.03MPa], indicating that the equation predicts a lower value at times t1 and t2, and a higher value at time t3. According to the exponential decay rule, for example, weights are set to give higher importance to the most recent data. Assuming the decay weight factors for the three times are 0.5, 0.7, and 0.9, corresponding to t1, t2, and t3 respectively, where t3 is the most recent time, the residual sequence is weighted. The weighted residuals are not directly used for comparison; instead, they are squared, multiplied by the corresponding weight, and then summed to form the value of the error accumulation function. Specifically, the weighted sum of squares is calculated to be 0.01. 2 ×0.5+0.01 2 ×0.7+(-0.03) 2 ×0.9=0.00093, which is the cumulative error measure of the current model on this data segment. The intake valve position sequence, as the control action input, directly affects the system dynamics. The actual measured value of the silo pump pressure sequence is the true response of the system, while the predicted value of the silo pump pressure is generated entirely by the equation containing the undetermined coefficient matrix based on the opening and flow data. The residual sequence reveals the deviation between the prediction and reality. The value of the error accumulation function, as the quantitative target, is continuously reduced through iteration, which means that the prediction is becoming more and more accurate. This allows the optimal intake valve position sequence to more accurately match the actual demand, avoid over-supply of air, fundamentally reduce the waste of compressed air, and achieve energy saving.

[0048] Step S2023: Derive the normal equation that makes its gradient zero based on the error accumulation function, and iteratively update the element values ​​of the undetermined coefficient matrix until the change in the coefficient matrix of adjacent iteration steps is less than the preset convergence threshold, and obtain the converged coefficient matrix.

[0049] The specific steps of step S2023 are as follows: Step S20231: Calculate the partial derivatives of the elements of the undetermined coefficient matrix using the error accumulation function to obtain the gradient vector expression composed of first-order partial derivatives. Set the gradient vector expression equal to the zero vector to establish a matrix equation with the undetermined coefficient matrix as an unknown quantity.

[0050] In this embodiment, the first-order partial derivative of the error accumulation function with respect to each element of the undetermined coefficient matrix is ​​calculated. All partial derivatives are arranged in order to form a gradient vector. The gradient vector represents the steepest rising direction of the error accumulation function value as the undetermined coefficient matrix changes. According to optimization theory, the partial derivatives of the function with respect to all independent variables approach zero at the minimum point. Let the gradient vector expression be equal to a zero vector, that is, the rate of change of the error accumulation function at the extreme point is zero. A series of equations with the elements of the undetermined coefficient matrix as common unknowns are obtained. These equations are combined to establish a matrix equation with the undetermined coefficient matrix as the only unknown. The accuracy of solving the matrix equation determines the minimum air pressure required to maintain the transmission, eliminates the reserved pressure margin, and avoids excessive energy supply and waste.

[0051] Step S20232: Decompose the undetermined coefficient matrix and vector terms in the matrix equation, and incrementally update them by a preset attenuation factor to obtain the normal equation. The vector terms are composed of the actual measured values ​​of the intake valve position sequence, intake flow rate sequence and silo pump pressure sequence.

[0052] In this embodiment, the undetermined coefficient matrix is ​​decomposed into the sum of the outer products of the previous time step estimate and the current data, and the vector terms are decomposed into the sum of the products of the previous time step estimate and the current data. A decay factor between 0 and 1 is preset, which determines the decay rate of the influence of historical data. The matrix equation is transformed into a recursive form that uses new data to incrementally update the original equation through the decay factor, resulting in the normal equation. The vector terms are a weighted combination of the actual measured values ​​of the intake valve position sequence, intake flow rate sequence, and silo pump pressure sequence, ensuring a direct correlation between the normal equation and the real physical process.

[0053] Step S20233: Substitute the undetermined coefficient matrix as the initial solution into the normal equation to calculate the residual vector and Jacobian matrix.

[0054] In this embodiment, the undetermined coefficient matrix is ​​used as the starting solution for the current iteration and substituted into the normal equation. The right-hand side of the normal equation is a vector term, and the left-hand side is the result vector obtained by multiplying the current undetermined coefficient matrix with the weighted data coefficient matrix. Subtracting the left-hand vector from the right-hand vector yields the residual vector under the current parameter solution. The residual vector reflects the degree of mismatch between the current parameter estimate and the ideal value. The second derivative information of the current error accumulation function with respect to the undetermined coefficient matrix is ​​calculated. Specifically, the derivative of each component in the gradient vector with respect to the elements in the undetermined coefficient matrix is ​​calculated. All derivatives are arranged into a square matrix to obtain the Jacobian matrix. The Jacobian matrix represents the local curvature of the gradient vector as it changes with the undetermined coefficient matrix. Based on the historical data of the intake valve position sequence and the intake flow sequence, a series of data vectors are combined. The outer product of these data vectors with their own transpose is calculated, and all these outer product matrices are accumulated in the time dimension. Combined with the exponential decay weight composed of decay factors, the weighted data coefficient matrix is ​​calculated. The weighted data coefficient matrix reflects the internal structural relationship between historical control actions and airflow state.

[0055] Step S20234: Establish a system of linear equations using the Jacobian matrix as coefficients and the residual vector as the right-hand side. Obtain the correction amount of the undetermined coefficient matrix by solving the system of linear equations, and iteratively update the undetermined coefficient matrix.

[0056] In this embodiment, a system of linear equations is constructed based on the Jacobian matrix and the residual vector, with the Jacobian matrix as the coefficient matrix and the residual vector as the right-hand side. This system describes the linear adjustment required by the undetermined coefficient matrix to reduce the residual vector to zero near the current parameter estimation point. The correction vector is obtained by solving the system of linear equations. The solution method uses matrix inversion. The correction vector is reorganized according to the order corresponding to the elements of the undetermined coefficient matrix to form the correction matrix. The undetermined coefficient matrix of the current iteration step is added to the correction matrix to obtain the updated undetermined coefficient matrix. Through the update operation, the pressure-flow equation can accurately represent the dynamic relationship between gas flow and pressure under the current ash and pipeline conditions, avoiding excessive supply or unstable delivery of compressed air due to inaccurate calculations.

[0057] Step S20235: Compare the L2 norm of the difference between the corresponding elements of the undetermined coefficient matrix of the current iteration step and the previous iteration step with the preset convergence threshold. If it is greater than the convergence threshold, continue the iteration update. If it is less than or equal to the convergence threshold, use the current undetermined coefficient matrix as the converged coefficient matrix.

[0058] In this embodiment, the difference between the undetermined coefficient matrix obtained in the current iteration step and the undetermined coefficient matrix used in the previous iteration step is calculated to evaluate whether the undetermined coefficient matrix is ​​close to the theoretical optimal value. The difference matrix is ​​obtained by subtracting the elements at all corresponding positions in the two matrices. The sum of squares of all elements in the difference matrix is ​​calculated, and the square root operation is performed on the sum of squares. The resulting scalar result is the L2 norm of the difference between the two matrices. The L2 norm is compared with a preset convergence threshold representing the allowable error range. If the L2 norm is greater than the convergence threshold, it indicates that the undetermined coefficient matrix has still changed significantly. It is used as the initial solution to continue iterating until it is less than or equal to the convergence threshold. If the L2 norm is less than or equal to the convergence threshold, it indicates that the undetermined coefficient matrix has reached a stable state. The undetermined coefficient matrix obtained in the current iteration step is used as the converged coefficient matrix required for the pressure-flow equation. The convergence threshold is an integer, and its size is set by those skilled in the art according to the actual situation.

[0059] Step S2024: Substitute the coefficient matrix after iterative convergence into the discrete-time difference equation to generate a pressure-flow equation in time-varying parameter form.

[0060] In this embodiment, the coefficient matrix after iterative convergence is substituted into the discrete-time difference equation. Specifically, the elements in the coefficient matrix are replaced with the undetermined coefficient symbols at the corresponding positions in the discrete-time difference equation, and instantiated into a pressure-flow equation in time-varying parameter form that can describe the dynamic characteristics of the current system. This equation represents the quantitative coupling relationship between the historical sequence of the intake valve position, the historical sequence of the intake flow rate, and the historical sequence of the silo pump pressure within the most recent time window, and quantifies the minimum supply pressure and flow rate required for delivery.

[0061] Step S203: Substitute the pipeline resistance coefficient as a correction factor into the time-varying parameter form of the pressure-flow equation, and establish a Jacobian matrix for the pipeline resistance coefficient to update the parameters in the time-varying parameter form of the pressure-flow equation, thus obtaining the pressure-flow equation.

[0062] In this embodiment, the pipeline resistance coefficient is multiplied by the parameters related to flow friction and inertia in the time-varying parameter form of the pressure-flow equation. This determines the pipeline operating condition information embedded in the time-varying parameter form of the pressure-flow equation. By analyzing the sensitivity of the output of the time-varying parameter form of the pressure-flow equation to the pipeline resistance coefficient, the first derivative vector of the output value of the time-varying parameter form of the pressure-flow equation under the current parameters is calculated as a function of the pipeline resistance coefficient. A matrix about the pipeline resistance coefficient is constructed. Using the sensitivity relationship determined by the matrix, combined with the deviation between the measured data and the prediction of the time-varying parameter form of the pressure-flow equation, the parameters in the time-varying parameter form of the pressure-flow equation are updated to obtain the final pressure-flow equation.

[0063] Step S204: Using the ash-to-gas ratio as the gain coefficient, construct a set of differential equations about the rate of change of the pressure of the silo pump and the rate of change of the inlet flow rate by simultaneously solving the pressure-flow equations.

[0064] In this embodiment, the ash-to-gas ratio, reflecting the energy efficiency of the conveying process, is substituted into the pressure-flow equation as a gain coefficient to establish a quantitative relationship between the change in inlet flow rate and the rate of mass reduction of ash plugs in the silo pump. Based on the principle of mass conservation, the rate of change of ash plug mass is proportional to the product of inlet flow rate and ash-to-gas ratio. The pipeline friction loss coefficient and pressure loss correction term are introduced, and combined with the mechanical relationship between the inlet valve position, inlet flow rate and silo pump pressure change rate in the pressure-flow equation, a set of differential equations is constructed with the silo pump pressure change rate and inlet flow rate change rate as common state variables. Through the differential equation set, gas consumption and solid conveying efficiency can be directly correlated. With the current ash-to-gas ratio as the target, the minimum required inlet flow rate is solved and tracked in real time while ensuring the stable movement of ash plugs, so as to complete the conveying task with the least gas consumption. The pipeline friction loss coefficient and pressure loss correction term are set by simulation experiments according to the pipeline material and ash characteristics.

[0065] Step S205: Using the current silo pump pressure, inlet valve position, and the initial solution state of the differential equation system as initial conditions, perform numerical integration on the differential equation system to generate the silo pump pressure change curve and gas consumption curve for ash hydrant transportation in the future finite time domain.

[0066] In this embodiment, the real-time measured pressure of the silo pump and the inlet valve position are used as the starting point to solve the steady-state initial values ​​of the differential equation system at this moment, which together constitute the initial conditions for numerical integration. Using the numerical integration method, starting from the current moment, within each integration step, based on the current pressure and flow state and the preset future inlet valve position control sequence, the instantaneous values ​​of the silo pump pressure change rate and the inlet flow rate change rate are calculated through the differential equation system. The silo pump pressure value and the inlet flow rate value of the next integration step are updated using the instantaneous values ​​of the silo pump pressure change rate and the inlet flow rate change rate. The calculation process is repeated, and the silo pump pressure values ​​calculated in each integration step within the future finite time domain are smoothly connected to generate the silo pump pressure change curve. The inlet flow rate value calculated in each step is accumulated over time to generate the cumulative gas consumption curve. The future inlet valve position control sequence is initialized based on the silo pump pressure, pipeline resistance coefficient and ash-to-gas ratio at the current moment, and dynamically adjusted during the transportation process.

[0067] Step S3: Based on the pressure change curve and gas consumption curve of the silo pump, establish a delivery cost function for the intake valve position increment and pressure tracking deviation, and solve it to obtain the optimal intake valve position sequence in the future finite time domain.

[0068] The specific steps of step S3 are as follows: Step S301: Based on the pressure change curve and air consumption curve of the silo pump, determine the pressure reference value and allowable air consumption threshold at different future times, and calculate the minimum safe pressure curve within the future limited time domain by combining the pipeline resistance coefficient, ash-to-air ratio and ash hopper level change data.

[0069] In this embodiment, based on the pressure change curve and gas consumption curve of the silo pump, the predicted pressure value corresponding to each future sampling time is extracted as a pressure reference, and the gas consumption at the corresponding time is extracted as a flow reference. According to the amount of ash to be transported indicated by the ash hopper level change data, the future ash demand is converted into the theoretical required gas volume using the ash-to-gas ratio. The pipeline resistance coefficient and the theoretical required gas volume are input into the pressure calculation formula based on two-phase fluid dynamics. The pressure calculation formula describes the minimum gas kinetic energy required to overcome specific pipeline resistance and drive the expected amount of ash plugs. Its output is the minimum pressure necessary to ensure the continuous and stable movement of ash plugs and avoid their stagnation or breakage. By solving the pressure calculation formula at each future sampling time and connecting the results in chronological order, the minimum safe pressure curve that changes with time is obtained. The reliability of the transport can be intuitively judged by the minimum safe pressure curve. The future sampling time is divided from the future finite time domain, and its size is determined by those skilled in the art based on the actual situation.

[0070] Step S302: Based on the minimum safe pressure curve and the theoretical minimum gas consumption, a pressure reference curve and an allowable gas consumption threshold curve are obtained through preset process safety margin compensation. The theoretical minimum gas consumption is calculated based on the ash-to-gas ratio and the ash conveying rate.

[0071] In this embodiment, by applying a process safety margin compensation based on the minimum safe pressure curve, instantaneous disturbances under uncertain operating conditions are addressed. The process safety margin compensation is dynamically adjusted according to the level of the pressure setpoint and the ease of conveying reflected by the ash-to-gas ratio. In the high-pressure section or when the ash-to-gas ratio is low, the process safety margin compensation increases; in the low-pressure section or when the ash-to-gas ratio is normal, the process safety margin compensation decreases. The process safety margin compensation is added point by point to the minimum safe pressure curve to generate a pressure reference curve. Based on the ash-to-gas ratio and the total amount of ash to be conveyed calculated from the ash hopper level change data, the theoretical minimum gas consumption to complete all conveying tasks is determined. The theoretical minimum gas consumption is multiplied by a safety factor, where the safety factor is greater than 1 and positively correlated with the pipeline resistance coefficient, to obtain the upper limit of the maximum allowable cumulative gas consumption within a finite time domain. The upper limit of the maximum cumulative gas consumption is evenly distributed over time and combined with the predicted conveying stage division to generate an allowable gas consumption threshold curve that changes over time. Through the pressure reference curve and the allowable gas consumption threshold curve, adaptive constraint planning is performed on pressure and gas consumption.

[0072] Step S303: Subtract the pressure reference curve from the pressure change curve of the silo pump point by point to obtain the pressure tracking deviation sequence, initialize the intake valve position increment sequence as the decision vector, and construct the delivery cost function based on the pressure tracking deviation sequence and the decision vector.

[0073] In this embodiment, the pressure reference curve and the predicted pressure change curve of the silo pump are subtracted point by point at the same time in the future, and arranged in chronological order to form a pressure tracking deviation sequence. The intake valve position increment sequence is initialized and used as the decision vector to be optimized. Each element of the decision vector represents the amount of change in the intake valve position relative to its previous set value. Based on the decision vector and the pressure tracking deviation sequence, a delivery cost function is constructed by combining a weighting coefficient used to balance the pressure tracking accuracy and valve action energy consumption. The smaller the value of the delivery cost function, the smoother the valve action and the less air consumption while meeting the pressure tracking requirements. The values ​​of the intake valve position increment sequence are a set of values ​​that minimize the delivery cost function under the conditions of the minimum safe pressure curve, the upper limit of total air consumption, and the intake valve position data. The weighting coefficients are set to positive values, and their magnitudes are set by those skilled in the art according to the actual situation.

[0074] Step S304: The minimum safe pressure curve and the total gas consumption upper limit obtained by integrating the allowable gas consumption threshold curve are used as boundary conditions and transformed into a set of linear inequalities to form a set of linear constraints in the future finite time domain.

[0075] In this embodiment, the boundary conditions are transformed into linear constraints on the decision vector. The minimum safe pressure curve is used as the pressure safety constraint, and the upper limit of total gas consumption is used as the total energy consumption constraint. The pressure safety constraint is represented as a linear inequality at each future sampling time, i.e., the predicted pressure value is greater than or equal to the minimum safe pressure value. The total energy consumption constraint is the sum of the predicted gas consumption at all times within the future finite time domain, and the sum does not exceed the upper limit of total gas consumption determined by the allowable gas consumption threshold curve. The total energy consumption constraint is achieved by imposing a restriction on the linear combination of the decision vector. Combined with the physical stroke and rate limit of the intake valve, it is transformed into a linear inequality about the intake valve position increment sequence and its cumulative value. All linear inequalities together constitute a complete set of linear constraints, which defines the feasible region range that the decision vector must follow during the optimization search process.

[0076] Step S305: Calculate the impact of the decision vector on the value of the transport cost function within the feasible region defined by the linear constraint set, and take the value of the decision vector that satisfies the linear constraint set and minimizes the value of the transport cost function as the optimal intake valve position sequence in the future finite time domain.

[0077] The specific steps of step S305 are as follows: Step S3051: Determine the feasible region of the decision vector based on the set of linear constraints, and use the current intake valve position as the initial point for iterative search.

[0078] In this embodiment, the allowable range of values ​​for the decision vector is determined based on the linear inequalities in the set of linear constraints. Each linear inequality defines an upper or lower limit for the decision vector. The set of decision vector values ​​defined by all linear inequalities constitutes the feasible region of the iterative search. The valve position value of the intake valve at the current moment is set as the reference valve position for the first control cycle, and the valve position increments of all subsequent cycles are set to zero, generating a zero valve position increment sequence. The zero valve position increment sequence constitutes the values ​​of the decision vector within the feasible region, which is used as the initial point within the feasible region. Here, the control cycle is represented as a fixed time period for solving and issuing energy-saving control commands.

[0079] Step S3052: Within the feasible region, the gradient information of the decision vector is obtained by taking the partial derivative of the cost function with respect to the decision vector at the current iteration point. Based on this, a direction vector pointing in the descent direction and within the feasible region is constructed.

[0080] In this embodiment, at the current iteration point, the partial derivatives of the transport cost function with respect to each element of the decision vector are calculated. By analyzing these partial derivatives, the rate of change of the pressure tracking deviation term in the transport cost function can be analyzed when the valve position increment changes slightly. The gradient vector formed by all partial derivatives represents the feasible direction in which the value of the transport cost function decreases the fastest. The gradient vector is then corrected by identifying the constraints that restrict the decision vector to the boundary at the current iteration point and calculating their boundary normal vectors. All boundary normal vectors are used as a set of basis vectors to span a linear subspace. The gradient vectors are orthogonally projected onto the orthogonal complement of the linear subspace. The projection matrix is ​​constructed using the boundary normal vectors, eliminating the components in the gradient that point out of the constraint boundary, and obtaining a direction vector pointing to the descent direction of the transport cost function and moving along the interior of the feasible region.

[0081] Step S3053: Update the value of the decision vector along the direction vector at the initial point according to the preset iteration step size, and calculate the pressure tracking deviation sequence corresponding to the updated decision vector.

[0082] In this embodiment, the decision vector value of the current iteration point is added to the result of multiplying the direction vector by a preset iteration step size to obtain a set of updated decision vector values. Based on the updated decision vector values ​​and the reference valve position of the current intake valve, the intake valve position proposal sequence is derived. The intake valve position proposal sequence is input into the differential equation system, and the corresponding chamber pump pressure prediction curve is calculated through numerical integration. The chamber pump pressure prediction curve is compared with the pressure reference curve to generate a new pressure tracking deviation sequence. The updated decision vector value itself is the updated intake valve position increment sequence. The iteration step size is set by those skilled in the art according to the actual situation. The iteration step size ensures that the new point after the movement is still within the feasible region and can effectively reduce the delivery cost function.

[0083] Step S3054: Substitute the updated pressure tracking deviation sequence into the transport cost function, calculate the transport cost function value, and iteratively update the position of the decision vector within the feasible region.

[0084] In this embodiment, the updated pressure tracking deviation sequence is substituted into the transport cost function to calculate the sum of squares of all elements in the pressure tracking deviation sequence and the sum of squares of all elements in the decision vector. The sum is then weighted using weighting coefficients to obtain a scalar value representing the overall cost, i.e., the transport cost function value. The current transport cost function value is compared with the transport cost function value of the previous iteration point to determine whether descent along the direction vector has been achieved. If descent occurs, the current decision variable value is taken as the current iteration point, and the next round of decision variable position updates begins until the transport cost function value reaches its minimum value. If descent does not occur, the decision variable of the previous iteration point remains unchanged, and updates are performed by adjusting the iteration step size or search direction.

[0085] Step S3055: Take the value of the decision vector that minimizes the transport cost function and satisfies all linear constraints, and convert it into the optimal intake valve position sequence in the future finite time domain.

[0086] In this embodiment, the search stops when the iteration meets the convergence condition. The convergence condition includes the decrease in the value of the transport cost function or the change in the decision variable being less than the corresponding change threshold in multiple consecutive iterations. The current value of the decision variable is taken as the solution that minimizes the transport cost function in the feasible region. This solution represents the optimal balance between transport stability and valve operation energy consumption. The decision variable is accumulated with the actual valve position of the intake valve at the current moment to derive the absolute valve position setting value corresponding to each future control cycle. The absolute valve position setting values ​​are combined in chronological order to generate the optimal intake valve position sequence. The optimal intake valve position sequence can ensure the continuous and stable movement of the ash plug and prevent the risk of pipe blockage while matching the minimum power required for ash plug transport, thereby reducing excessive consumption of compressed air and ineffective valve operation, and reducing the unit energy consumption of the ash transport cycle.

[0087] Step S4: Send the optimal intake valve position sequence to the intake valve for execution and adjust the corresponding silo pump. Based on the pressure deviation optimization command during the execution process, perform energy-saving control on the ash conveying pipeline network.

[0088] The specific steps of step S4 are as follows: Step S401: Using the minimum pressure required to maintain stable ash hydrant delivery as a reference, convert the intake valve position value in the optimal intake valve position sequence into an energy-saving control command to drive the intake valve.

[0089] In this embodiment, the dynamic pressure threshold determined by the minimum safe pressure curve is used as a benchmark to perform the conversion and generation of control commands. The intake valve position value in the optimal intake valve position sequence is correlated and mapped with the minimum pressure value necessary to maintain the movement of the ash plug at the current moment, generating corresponding drive commands. The correlation mapping process ensures that the core objective of the drive command output is to make the silo pump pressure track the minimum benchmark and limit the lower limit of the air supply pressure. During the correlation mapping, the valve position value is smoothed, the rate of change between adjacent valve position values ​​is calculated, and the portion exceeding the preset maximum rate of change is limited, generating a valve position command curve with a gradual change. The smoothing process reduces the frequent large-amplitude movement of the valve, reducing the additional compressed air consumption and equipment wear caused by abnormal movement. The valve position command curve is converted into a standard control signal for driving the intake valve positioner in chronological order. The standard control signal sequence constitutes an energy-saving control command with the core feature of minimizing power supply. The maximum rate of change is set by those skilled in the art according to the actual situation.

[0090] Step S402: Send the gas consumption curve and the pressure change curve of the silo pump within the future finite time domain as feedforward values ​​to the silo pump to adjust its output power.

[0091] In this embodiment, the changing trend and rate of compressed air demand are obtained by analyzing the air consumption curve, and the dynamic trajectory and gradient of the pressure setpoint are obtained by analyzing the pressure change curve of the silo pump. The changing trend and rate, dynamic trajectory and gradient are converted into demand expectation instructions. The demand expectation instructions specify the pressure and compressed air flow required to match the ash hydrant conveying process. The demand expectation instructions are sent to the controller that provides power to the silo pump, and its output power setpoint is adjusted, thereby eliminating pressure fluctuations and flow overshoot caused by response lag.

[0092] Step S403: Obtain the current actual pressure inside the silo pump, and generate an instantaneous pressure deviation signal reflecting the movement status of the ash plug and the pipeline resistance by combining the execution result of the energy-saving control command of the air intake valve.

[0093] In this embodiment, the actual measured value of the pressure sensor inside the silo pump is collected and filtered to eliminate high-frequency noise, resulting in a stable actual pressure value of the silo pump. The execution result of the energy-saving control command by the intake valve is monitored, and the actual pressure value of the silo pump is compared with the expected pressure state corresponding to the currently executed energy-saving control command. The expected pressure state is calculated based on the currently executed energy-saving control command, the pipeline resistance coefficient, and the ash-to-gas ratio, representing the pressure level that should be achieved under ideal operating conditions. The actual pressure value is subtracted from the expected pressure value to obtain the instantaneous pressure difference. If the instantaneous pressure difference is consistently positive, it indicates that the actual resistance is higher than expected, which may indicate slow movement of ash plugs or partial blockage of the pipeline. If the instantaneous pressure difference is consistently negative, it indicates that the actual operating conditions are better than expected, and there is an energy-saving space due to excessive gas supply pressure. The sign and trend of the instantaneous pressure difference are analyzed. The potential is analyzed and converted into an instantaneous pressure deviation signal that reflects the degree of pressure deviation, the smoothness of instantaneous ash hydrant delivery, and the true state of pipeline resistance. The conversion process includes calculating the instantaneous sign, absolute amplitude, and linear fitting slope of the instantaneous pressure difference. If the instantaneous sign is positive and the linear fitting slope is also positive, it is determined that resistance is accumulating, and the output composite signal value is positive, and its magnitude is proportional to the weighted sum of the absolute amplitude and the linear fitting slope. If the instantaneous sign is negative and the linear fitting slope is also negative, it is determined that there is excess power and it is continuous, and the output composite signal value is negative, and its magnitude is proportional to the weighted sum of the absolute value of the absolute amplitude and the absolute value of the linear fitting slope. In other cases, such as when the instantaneous sign and the linear fitting slope are opposite and the amplitude is extremely small, the output composite signal is close to zero, indicating that the state is stable. Finally, the output value of the instantaneous pressure deviation signal is obtained.

[0094] In this example, assuming sampling is performed once per second, within a 5-second analysis window, the instantaneous pressure deviation signal sequence is generated as [-0.002, +0.005, +0.015, +0.030, +0.048], in megapascals (MPa). The arithmetic mean is +0.0192 MPa, indicating that the pressure remains higher than expected. The variance of the instantaneous pressure deviation signal sequence is 0.00038. Linear fitting of these five points yields a slope representing the trend, +0.0125 MPa / s, indicating that the deviation is progressing rapidly. The rate increases linearly. Since the mean and slope are both positive, it is judged as a medium-to-high risk condition where pipeline resistance is increasing. Assuming that the risk level is determined by the weighted sum of the mean and 0.5 times the slope, the calculated value is 0.0192 + 0.5 × 0.0125 = 0.02545, which is judged as medium risk. The corresponding control action is to increase the intake valve position value of the next 3 control cycles by 3% based on the original optimal sequence. Assuming that the original optimal intake valve opening sequence for the 1st, 2nd, and 3rd seconds is [62%, 60%, 58%], it is corrected to [65%, 63%, 61%] according to the adjustment instruction.

[0095] Step S404: Determine the current conveying efficiency and air consumption status of the ash hydrant based on the instantaneous pressure deviation signal, and adjust the air inlet valve position value to achieve energy-saving ash conveying control.

[0096] In this embodiment, the temporal characteristics of the instantaneous pressure deviation signal are analyzed to determine the actual efficiency of the current ash plug conveying and the compressed air consumption status. When the instantaneous pressure deviation signal continuously fluctuates around zero, it indicates that the ash-to-air ratio is stable and the conveying is in a high-efficiency stage. When the instantaneous pressure deviation signal shows a positive increasing trend, it indicates that the pipeline resistance is increasing, the ash plug moving speed is decreasing, the current conveying efficiency is decreasing, and there is a risk of pipe blockage. According to the rate and magnitude of the deviation increase, the energy-saving control command to be executed is adjusted upward according to preset rules to increase the air supply power to overcome the increased resistance and avoid energy waste and production interruption caused by conveying failure. The force deviation signal shows a continuous negative increasing trend, indicating that the actual required power is lower than the prediction and the current air supply is excessive. According to the deviation adjustment energy control command, unnecessary compressed air supply is reduced, and redundant air supply is compressed in real time while ensuring reliability, so as to achieve energy-saving control of the ash conveying process. Among them, the preset rules are set by analyzing the instantaneous pressure deviation signal and encoded into condition judgment and function mapping relationship, including the pre-adjustment rule for increased resistance, which classifies the risk into different levels according to the weighted sum of the deviation value and the change trend value, and the power excess reduction rule, which calculates the air supply margin for reduction according to the magnitude of the deviation.

[0097] exist Figure 3 In the diagram, arrows represent airflow, circles represent material particles, and A, B, C, and D represent four points on the material particles. When the airflow passes around point A of the material particle, it changes direction, and the velocity drops instantaneously to 0. The air particles form a stagnation zone at point A. From point A, the airflow moves along the upper and lower surfaces of the material particles, i.e., the air particles move along AB or AC. Initially, the airflow adheres closely to the particle surface and moves along the curvature of the particle's surface. At this point, the flow is in the laminar stage of the boundary layer, where the boundary layer adheres closely to the surface of the solid particles, and the velocity gradually recovers from 0 to... When the airflow reaches point D along the particle surface, a thin layer of incoming velocity occurs, and boundary layer separation takes place. Due to the adverse pressure gradient on the particle surface, the kinetic energy of the airflow within the boundary layer is consumed, and it cannot maintain its attachment state. Therefore, it detaches from the particle surface. Point D is the starting point of the separation between the upper and lower surfaces. After the airflow separates from point D, a wake region is formed behind the particle. The multiple closed vortices on the right side of the particle are turbulent vortex structures formed by the separated flow. These vortices will continue to be generated and detached, and are the core area of ​​energy dissipation in the wake, as well as the main source of form drag.

[0098] Example 2: Please see Figure 4One embodiment of the present invention is a high-end automatic ash conveying big data energy-saving control system, which includes a working condition sensing module, an energy efficiency measurement module, a conveying optimization module, and an execution module.

[0099] The operating condition sensing module is used to collect upstream production data, ash conveying process data and downstream power data of the ash conveying pipeline network through visual detection, and calculate the pipeline resistance coefficient and ash-to-gas ratio through soft measurement estimation. The upstream production data includes ash hopper level change data.

[0100] The energy efficiency measurement module is used to input the pipeline resistance coefficient and ash-to-gas ratio into the pressure-flow equation to generate the pressure change curve and gas consumption curve of the silo pump to maintain ash plug delivery within a finite time domain in the future. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure, and inlet flow rate.

[0101] The delivery optimization module is used to establish a delivery cost function for the increment of the intake valve position and the pressure tracking deviation based on the pressure change curve and the gas consumption curve of the silo pump, and solve it to obtain the optimal intake valve position sequence in the future finite time domain.

[0102] The execution module is used to send the optimal intake valve position sequence to the intake valve for execution and adjust the corresponding silo pump. Based on the pressure deviation optimization command during the execution process, it performs energy-saving control on the ash conveying pipeline network.

[0103] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0104] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A high-end automatic ash conveying big data energy-saving control method, characterized in that, include: Upstream production data, ash conveying process data, and downstream power data of the ash conveying pipeline network are collected by visual inspection. The pipeline resistance coefficient and ash-to-gas ratio are calculated by soft measurement estimation. The upstream production data includes ash hopper level change data. By inputting the pipeline resistance coefficient and the ash-to-gas ratio into the pressure-flow equation, the pressure change curve of the silo pump and the gas consumption curve for maintaining the ash plug delivery within a future finite time domain are generated. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure and inlet flow. Based on the pressure change curve and gas consumption curve of the silo pump, a transport cost function for the increment of the inlet valve position and the pressure tracking deviation is established, and the optimal inlet valve position sequence in the future finite time domain is obtained by solving it. The optimal intake valve position sequence is sent to the intake valve for execution and the corresponding silo pump is adjusted. Based on the pressure deviation during the execution process, the optimization command is used to carry out energy-saving control of the ash conveying pipeline network.

2. The high-end automatic ash conveying big data energy-saving control method according to claim 1, characterized in that, The process involves collecting upstream production data, ash conveying process data, and downstream power data of the ash conveying pipeline network through visual inspection. Soft sensing estimation is used to calculate the pipeline resistance coefficient and ash-to-gas ratio. The upstream production data includes ash hopper level change data, including: The surface image sequence of the material in the ash hopper is collected by industrial vision sensors from upstream production data. The grayscale and texture features are identified by visual inspection to establish a three-dimensional contour of the material surface, and the material level change data of the ash hopper is measured based on the three-dimensional contour of the material surface. The upstream production data, ash conveying process data, and downstream power data are time-series aligned, and outlier data points are removed to obtain the conveying sample set; Based on the transport sample set, the pressure stability characteristics of the silo pump during the feeding stage, the pressure decay gradient during the transport stage, and the valve position time sequence integral value of the air inlet valve are calculated. Combined with the pipeline resistance coefficient as an unknown, a pressure decay differential equation is constructed. The pipeline resistance coefficient is obtained based on the pipeline geometry and state information. The root mean square error is calculated based on the pressure prediction value obtained by solving the pressure attenuation differential equation and the pressure change parameters in the ash conveying process data. The pipeline resistance coefficient is then adjusted iteratively using gradient descent until the root mean square error is less than the preset error threshold, thus obtaining the optimized pipeline resistance coefficient. The optimized pipeline resistance coefficient is substituted into the soft sensor estimation equation to calculate the ash-to-gas ratio of the current conveying cycle. The soft sensor estimation equation is derived based on the preset material conservation equation and gas state equation.

3. The high-end automatic ash conveying big data energy-saving control method according to claim 2, characterized in that, The calculation of the pressure stability characteristics during the pump feeding stage, the pressure attenuation gradient during the conveying stage, and the valve position time-series integral value of the intake valve based on the conveying sample set includes: Based on the pressure time series data of the silo pump during the feeding stage extracted from the transport sample set, its pressure variance was calculated as a pressure stationary feature. Based on the transport sample set, the pressure of the silo pump is identified to enter the monotonically decreasing transport stage. Linear piecewise fitting is performed on the decreasing stage, and the resulting slope is used as the pressure decay gradient. The timing data of the valve position command and feedback signal of the intake valve are extracted from the transported sample set, and the corresponding valve position timing integral value is obtained by numerical integration.

4. The high-end automatic ash conveying big data energy-saving control method according to claim 3, characterized in that, The process involves inputting the pipeline resistance coefficient and ash-to-gas ratio into the pressure-flow equation to generate the pressure change curve and gas consumption curve of the silo pump for maintaining ash plug delivery within a finite time domain. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure, and inlet flow rate, and includes: The conveying sample set is resampled according to a preset time step, taking the conveying cycle of the ash conveying pipeline as the unit, to generate the inlet valve position sequence, the silo pump pressure sequence, and the inlet flow rate sequence. The intake valve position sequence, the silo pump pressure sequence, and the intake flow rate sequence are used as inputs to a preset discrete-time difference equation containing an undetermined coefficient matrix. The pressure-flow equation in time-varying parameter form is determined by recursively updating the undetermined coefficient matrix. The pressure-flow equation in time-varying parameter form is obtained by substituting the pipe resistance coefficient as a correction factor into the pressure-flow equation and establishing a Jacobian matrix for the pipe resistance coefficient to update the parameters in the pressure-flow equation in time-varying parameter form. Using the ash-to-gas ratio as the gain coefficient, a set of differential equations is constructed by simultaneously solving the pressure-flow equations for the pressure change rate of the silo pump and the inlet flow rate change rate. Using the current silo pump pressure, inlet valve position, and the initial solution state of the differential equation system as initial conditions, numerical integration of the differential equation system is performed to generate the silo pump pressure change curve and gas consumption curve for ash hydrant transportation within a finite time domain in the future.

5. The high-end automatic ash conveying big data energy-saving control method according to claim 4, characterized in that, The step of using the intake valve position sequence, the hopper pump pressure sequence, and the intake flow rate sequence as inputs to a preset discrete-time difference equation containing an undetermined coefficient matrix, and determining the time-varying parameter form of the pressure-flow equation by recursively updating the undetermined coefficient matrix, includes: Based on the intake valve position sequence, the chamber pump pressure sequence, and the intake flow rate sequence, a discrete-time difference equation containing an undetermined coefficient matrix is ​​constructed as the basic expression of the pressure-flow rate equation. The predicted output term of the silo pump pressure in the discrete-time difference equation is subtracted from the actual measured value of the silo pump pressure sequence, and the error accumulation function about the undetermined coefficient matrix is ​​established through the obtained residual sequence. The normal equation that makes its gradient zero is derived based on the error accumulation function, and the element values ​​of the undetermined coefficient matrix are iteratively updated until the change in the coefficient matrix of adjacent iteration steps is less than the preset convergence threshold, thus obtaining the converged coefficient matrix. Substituting the coefficient matrix after iterative convergence into the discrete-time difference equation, a pressure-flow equation in time-varying parameter form is generated.

6. The high-end automatic ash conveying big data energy-saving control method according to claim 5, characterized in that, The process involves deriving the normal equation whose gradient is zero based on the error accumulation function, and iteratively updating the element values ​​of the undetermined coefficient matrix until the change in the coefficient matrix in adjacent iterations is less than a preset convergence threshold, resulting in a converged coefficient matrix, including: By taking the partial derivatives of the elements of the undetermined coefficient matrix through the error accumulation function, we obtain the gradient vector expression composed of the first-order partial derivatives. By setting the gradient vector expression equal to the zero vector, we establish a matrix equation with the undetermined coefficient matrix as the unknown quantity. The matrix equation is decomposed into undetermined coefficient matrix and vector terms, and incrementally updated by a preset attenuation factor to obtain the normal equation. The vector terms are composed of the actual measured values ​​of the intake valve position sequence, intake flow sequence and silo pump pressure sequence. Substitute the undetermined coefficient matrix into the normal equation as the initial solution to calculate the residual vector and the Jacobian matrix. A system of linear equations is established using the Jacobian matrix as coefficients and the residual vector as the right-hand side. The correction amount of the undetermined coefficient matrix is ​​obtained by solving the system of linear equations, and the undetermined coefficient matrix is ​​iteratively updated. The L2 norm of the difference between the corresponding elements of the undetermined coefficient matrix of the current iteration step and the previous iteration step is compared with the preset convergence threshold. If it is greater than the convergence threshold, the iteration update continues. If it is less than or equal to the convergence threshold, the current undetermined coefficient matrix is ​​used as the converged coefficient matrix.

7. The high-end automatic ash conveying big data energy-saving control method according to claim 6, characterized in that, Based on the pressure change curve and gas consumption curve of the silo pump, a delivery cost function is established regarding the intake valve position increment and pressure tracking deviation. Solving this function yields the optimal intake valve position sequence within a future finite time domain, including: Based on the pressure change curve and gas consumption curve of the silo pump, determine the pressure reference value and allowable gas consumption threshold at different future times, and calculate the minimum safe pressure curve in the future limited time domain by combining the pipeline resistance coefficient, ash-to-gas ratio and ash hopper level change data. Based on the minimum safe pressure curve and the theoretical minimum gas consumption, a pressure reference curve and an allowable gas consumption threshold curve are obtained by compensating for the preset process safety margin. The theoretical minimum gas consumption is calculated based on the ash-to-gas ratio and the ash conveying rate. The pressure reference curve and the pressure change curve of the silo pump are subtracted point by point to obtain the pressure tracking deviation sequence. The intake valve position increment sequence is initialized as the decision vector. The delivery cost function is constructed based on the pressure tracking deviation sequence and the decision vector. The minimum safe pressure curve and the total gas consumption upper limit obtained by integrating the allowable gas consumption threshold curve are used as boundary conditions, which are transformed into a set of linear inequalities to form a set of linear constraints in the future finite time domain. The influence of the decision vector on the transport cost function value is calculated within the feasible region defined by the linear constraint set. The value of the decision vector that satisfies the linear constraint set and minimizes the transport cost function value is taken as the optimal intake valve position sequence in the future finite time domain.

8. The high-end automatic ash conveying big data energy-saving control method according to claim 7, characterized in that, The step of calculating the impact of the decision vector on the transport cost function value within the feasible region defined by the linear constraint set, and taking the value of the decision vector that satisfies the linear constraint set and minimizes the transport cost function value as the optimal intake valve position sequence in the future finite time domain, includes: The feasible region of the decision vector is determined based on the set of linear constraints, and the current intake valve position is used as the initial point for the iterative search. Within the feasible region, the gradient information of the decision vector is obtained by taking the partial derivative of the cost function with respect to the decision vector at the current iteration point. Based on this, a direction vector pointing in the descent direction and within the feasible region is constructed. The decision vector is updated along the direction vector at the initial point according to the preset iteration step size, and the pressure tracking deviation sequence corresponding to the updated decision vector is calculated. Substitute the updated pressure tracking deviation sequence into the transport cost function, calculate the transport cost function value, and iteratively update the position of the decision vector in the feasible region. For the decision vector that minimizes the transport cost function and satisfies all linear constraints, it is converted into the optimal intake valve position sequence in the future finite time domain.

9. The high-end automatic ash conveying big data energy-saving control method according to claim 8, characterized in that, The process of sending the optimal intake valve position sequence to the intake valve for execution and adjusting the corresponding silo pump, and optimizing the ash conveying pipeline network based on the pressure deviation during execution, includes: Using the minimum pressure required to maintain stable delivery of ash plugs as a benchmark, the intake valve position values ​​in the optimal intake valve position sequence are converted into energy-saving control commands to drive the intake valves. The gas consumption curve and the pressure change curve of the silo pump within a limited time domain in the future are sent as feedforward values ​​to the silo pump to adjust its output power. The actual pressure inside the current silo pump is obtained, and the instantaneous pressure deviation signal reflecting the movement state of the ash plug and the pipeline resistance is generated by combining the energy-saving control command execution result of the air intake valve. The current conveying efficiency and air consumption status of the ash hydrant are determined based on the instantaneous pressure deviation signal, and the air inlet valve position is adjusted to achieve energy-saving ash conveying control.

10. A high-end automatic ash conveying big data energy-saving control system, used to implement the high-end automatic ash conveying big data energy-saving control method according to any one of claims 1-9, characterized in that, It includes a working condition sensing module, an energy efficiency measurement module, a transport optimization module, and an execution module: The working condition perception module is used to collect upstream production data, ash conveying process data and downstream power data of the ash conveying pipeline network through visual detection, and calculate the pipeline resistance coefficient and ash-to-gas ratio through soft measurement estimation. The upstream production data includes ash hopper level change data. The energy efficiency measurement module is used to input the pipeline resistance coefficient and ash-to-gas ratio into the pressure-flow equation to generate the pressure change curve and gas consumption curve of the silo pump to maintain ash plug delivery within a future finite time domain. The pressure-flow equation is constructed based on the inlet valve position, silo pump pressure and inlet flow. The delivery optimization module is used to establish a delivery cost function for the inlet valve position increment and pressure tracking deviation based on the pressure change curve and gas consumption curve of the silo pump, and solve it to obtain the optimal inlet valve position sequence in the future finite time domain. The execution module is used to send the optimal intake valve position sequence to the intake valve for execution and adjust the corresponding silo pump. Based on the pressure deviation optimization command during the execution process, it performs energy-saving control on the ash conveying pipeline network.

Images