Bulk material slow loading and fast unloading method

Abstract

This invention discloses a method for slow loading and fast unloading of bulk materials, relating to the field of bulk material conveying and unloading technology. The invention uses multi-source sensors to collect real-time operational data from a horizontal continuous loader and downstream loading equipment, establishing a transportation dataset and a cargo dataset, and performing standardization processing. Based on the standardized data, it extracts transportation operation feature vectors and cargo state feature vectors. According to the transportation operation feature vectors, it sets up traffic flow theory and a congestion wave propagation model to generate joint control result one. Based on the cargo state feature vectors, it performs material concentration field modeling and inversion analysis to generate joint control result two. The two joint control results are fused and optimized to obtain the final joint control result, enabling closed-loop control of the loader and loading equipment, followed by anomaly identification and deep source tracing. This invention achieves coordinated control of the loading and unloading processes, solves the material compaction problem caused by slow loading, and improves the thoroughness and efficiency of unloading.

Description

Technical Field

[0001] This invention relates to the field of bulk material conveying and unloading technology, specifically a method for slow loading and fast unloading of bulk materials. Background Technology

[0002] Horizontal continuous loading operations are widely used in the loading, unloading, and transshipment of bulk materials such as coal, ore, grain, and cement. These operations typically require maintaining stable material flow during loading to reduce impact and dust, and achieving rapid discharge during unloading to improve transportation efficiency. Therefore, the so-called "slow loading, fast unloading" operation method has gradually emerged in engineering practice.

[0003] In existing technologies, slow loading is generally achieved by adjusting the operating parameters of the conveying equipment. For example, by using frequency converters to control the output flow of belt conveyors, screw conveyors, or vibrating feeders, materials are continuously fed into the loading space, specifically a car, hopper, or bin, at a low rate. In this process, the materials mainly rely on gravity to naturally accumulate and form a stockpile structure, and the loading process does not actively control the material accumulation pattern.

[0004] During the unloading stage, existing technologies mostly employ quick-opening unloading structures to achieve rapid unloading, such as quick-opening bottom gate structures, flap unloading structures, or overall tilting unloading devices. These methods release a large unloading cross-section at once, allowing the material to flow out quickly under gravity, thus shortening the unloading time.

[0005] However, the aforementioned slow loading and fast unloading methods are mostly simple combinations of loading control and unloading structures, and they still have obvious limitations in the continuous loading and unloading of bulk materials.

[0006] First, under slow and continuous loading conditions, the material gradually accumulates in the loading space and produces a compaction effect, which easily forms an uneven internal density accumulation structure, leading to increased friction between material layers and improved local structural stability, thus reducing the overall fluidity of the material.

[0007] Secondly, when using a quick-opening unloading channel for unloading, the internal structure of the stockpile has become dense, which can easily lead to bridging effects, flow deviation, or corner retention, resulting in residual material during the unloading process, affecting the thoroughness of unloading, and even requiring auxiliary vibration or manual intervention.

[0008] Furthermore, existing technologies typically employ independent control strategies for loading and unloading processes, lacking a collaborative adjustment mechanism based on the material accumulation state. They cannot perform feedforward optimization of the loading configuration according to unloading requirements, nor can they dynamically adjust the unloading strategy based on the stockpile structure during the unloading phase. Consequently, it is difficult to balance loading stability, unloading efficiency, and system reliability.

[0009] On the other hand, different bulk materials vary in particle size distribution, moisture content and adhesion characteristics. Existing slow loading and fast unloading systems mostly adopt fixed structures and fixed control logic, which have limited adaptability to various types of materials, further restricting their application effect under complex working conditions.

[0010] Therefore, how to actively control the material accumulation pattern during horizontal continuous loading and establish a collaborative control mechanism for loading and unloading processes to improve the thoroughness and stability of rapid unloading has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0011] The purpose of this invention is to solve the problem that in the existing slow loading and fast unloading process of bulk materials, the lack of a coordinated control mechanism for loading and unloading leads to the gradual compaction of materials during the slow loading stage, forming a dense and uneven accumulation structure, and then problems such as bridging, flow deviation and corner residue appearing during the fast unloading stage. Therefore, this invention proposes a slow loading and fast unloading method for bulk materials.

[0012] The objective of this invention can be achieved through the following technical solution: a method for slow loading and fast unloading of bulk materials, comprising the following steps:

[0013] S1: Real-time data collection of the operation of the horizontal continuous loader and its downstream loading equipment is carried out through multi-source sensors. Transportation dataset and cargo dataset are established respectively. The collected raw data is cleaned and standardized to obtain standardized transportation dataset and standardized cargo dataset.

[0014] S2: Based on the standardized transportation dataset and the standardized cargo dataset, extract the transportation operation feature vector and the cargo status feature vector respectively;

[0015] S3: Based on the characteristic vector of transportation operation, traffic flow theory and congestion wave propagation model are set up to predict and regulate the flow state of materials on the horizontal continuous loader and generate joint control result one.

[0016] S4: Based on the cargo loading state feature vector, the material distribution inside the loading equipment is inverted through material concentration field modeling and diffusion potential energy analysis to generate joint control result two;

[0017] S5: Perform phase alignment, weight allocation, and game-theoretic fusion optimization on joint control result one and joint control result two to obtain the final joint control result;

[0018] S6: Based on the final joint control results, perform closed-loop control on the horizontal continuous loader and loading equipment, monitor the control response in real time, identify and trace anomalies, and dynamically adjust the control strategy.

[0019] As a preferred embodiment of the present invention, the specific process for extracting the transportation operation feature vector and the cargo loading status feature vector is as follows:

[0020] By establishing sliding time windows with fixed time lengths for standardized transportation datasets and standardized cargo datasets, and sliding each window forward with a fixed step size, the continuous time series is divided into multiple analysis intervals.

[0021] Within each time window, statistical calculations are performed on the conveyor belt speed, instantaneous material flow rate, drive motor current, and conveying tension in the transportation dataset to obtain the transportation operation feature vectors.

[0022] Within the same time window, differential and statistical operations are performed on the load weight, material height, and loading posture tilt angle in the cargo dataset to obtain the cargo state feature vectors.

[0023] As a preferred embodiment of the present invention, the specific process for generating joint control result one includes:

[0024] The average conveying speed and average material flow rate are obtained from the transportation operation feature vector, and a material linear density function is established.

[0025] The continuous conveying process within the time window is divided into multiple micro-time periods. The mass of each micro-unit and its spatial position on the conveyor belt are calculated to form a micro-unit sequence.

[0026] Based on the micro-unit sequence, the spatial gradient of adjacent micro-units is calculated, and local congestion areas are identified according to traffic flow theory to calculate the congestion wave propagation speed.

[0027] Based on the congestion wave propagation speed and spatial gradient, the propagation distance of the accumulation wave in the future time is predicted; if the accumulation wave propagation distance reaches or exceeds the inlet distance of the loading equipment, the flow rate mitigation factor is calculated, and the micro-unit release time is adjusted accordingly; the adjusted micro-unit mass and release time are combined to form the joint control result one.

[0028] As a preferred embodiment of the present invention, the specific process for generating joint control result two includes:

[0029] Material level height data is obtained from the feature vector of the loading status. The space of the loading equipment is divided into a two-dimensional grid. The material mass of each grid is calculated based on the stacking height and unit volume density, and a material concentration field at the loading end is established.

[0030] The material diffusion coefficient is calculated based on the material characteristic parameters, and the Laplace operator is calculated for the material concentration field at the loading end to obtain the diffusion potential energy. The diffusion potential energy is then mapped to the local stacking risk function.

[0031] A mapping function between micro-units and loading end meshes is established to map each micro-unit in the micro-unit sequence to the local stacking risk value of the corresponding mesh, and the micro-unit release time is adjusted accordingly to form joint control result two.

[0032] As a preferred embodiment of the present invention, the specific process for obtaining the final joint control result includes:

[0033] Phase calibration is performed on joint control result one and joint control result two. Time delay correction is performed on the second joint control result based on the average material settling time in the cargo state characteristic vector to obtain a time-aligned control pair.

[0034] Calculate the control difference vector and obtain the weight coefficients of the second joint control result based on the control entropy value. The more unstable the system, the greater the weight of the first joint control result.

[0035] Establish a control cost function, and obtain the final joint control result by minimizing the control cost function; perform boundary correction on the final joint control result.

[0036] As a preferred embodiment of the present invention, the process of performing closed-loop control of the horizontal continuous loader and loading equipment based on the final joint control result is as follows:

[0037] Based on the final joint control results, a timing control vector is generated, and the device response vector is collected in real time to form a control response mapping sequence;

[0038] Based on the control response mapping sequence, steps S2 to S5 are re-executed to obtain the twin prediction output and calculate the prediction error;

[0039] An error evolution gradient is established based on the prediction error time series, and an anomaly discrimination function is established. When the value of the anomaly discrimination function exceeds the dynamic threshold, it is determined as an anomaly candidate point, and the error value at the corresponding time is recorded.

[0040] Based on the candidate anomalies, a control response coupling perturbation matrix is ​​established, the contribution of anomaly sources is calculated, and the set of dominant perturbation sources is extracted.

[0041] Based on the abnormal error value, the contribution of the abnormal source, and the control offset, a comprehensive grading function is established to classify the abnormality level;

[0042] Establish a processing cost function. If the processing cost exceeds the preset maximum allowable cost, the exception will be incorporated into the joint control to recalculate the process and trigger the corresponding loading facility adjustment action. Otherwise, the twin model parameters will be corrected.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] This invention addresses the problem of material compaction leading to unloading difficulties during slow, continuous loading by using feedforward control of material release based on a loading state feature vector. During loading, real-time data on the loading equipment's load weight, material level, and tilt angle are collected to establish a loading state feature vector, and the material accumulation morphology inside the loading equipment is analyzed. If a localized accumulation risk occurs, the release time of the material micro-units is adjusted accordingly, allowing subsequent material to actively avoid high-risk areas, resulting in a loose and uniform accumulation structure during loading. Using this method, the fluctuation in material density within the loading space is reduced, and the frictional resistance between material layers is decreased. During the rapid unloading phase, because the material pile does not form a dense structure, the material can flow smoothly after the unloading channel is opened, resulting in less residual material after unloading and eliminating the need for auxiliary vibration or manual cleaning. This proactive control of the material accumulation morphology during the loading phase successfully achieves the prediction and optimization of the unloading effect. Attached Figure Description

[0045] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0046] Figure 1 This is a schematic diagram of the principle of the present invention. Detailed Implementation

[0047] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0048] It should be understood that the terms “comprising” and “including” used in this disclosure and claims indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0049] It should also be understood that the terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the disclosure. As used in this disclosure and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this disclosure and claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations.

[0050] Please see Figure 1 As shown, a method for slow loading and fast unloading of bulk materials includes:

[0051] S1: Collect real-time data from the horizontal continuous loader and the loading equipment downstream of the horizontal continuous loader through corresponding sensors, and classify them into transportation datasets and cargo datasets; and perform data cleaning to obtain standardized transportation datasets and standardized cargo datasets.

[0052] S2: Based on the standardized transportation dataset and the standardized cargo dataset, feature extraction is performed to obtain the transportation operation feature vector and the cargo status feature vector.

[0053] S3: Based on the transportation operation characteristic vector, perform correlation analysis on the horizontal continuous loader and loading equipment to obtain joint control result one;

[0054] S4: Based on the cargo state feature vector, the loading equipment and the horizontal continuous loader are subjected to inverse correlation analysis to obtain joint control result two;

[0055] S5: Analyze joint control result one and joint control result two to obtain the final joint control result;

[0056] S6: Based on the final joint control results, control the horizontal continuous loader and loading equipment, and collect real-time data; execute the process from S2 to S5 to perform anomaly analysis to obtain anomaly points; conduct in-depth mining based on the anomaly points to obtain the set of dominant disturbance sources, classify the dominant disturbance sources according to their levels, and determine whether it is necessary to substitute them into the calculation or execute the corresponding facilities for processing.

[0057] The real-time acquisition of operational data through multi-source sensors installed on the horizontal continuous loader and its downstream loading equipment includes:

[0058] The data collected include conveyor belt speed v(t), conveyor motor current I(t), material flow rate q(t), conveyor tension T(t), loading equipment weight W(t), material height h(t), loading tilt angle θ(t), and loading space utilization η(t). Specifically, the data collected from the horizontal continuous loader side include conveyor belt speed v(t), instantaneous material flow rate q(t), drive motor current I(t), and conveyor tension T(t), forming the transportation dataset D. T The loading equipment collects data such as the load weight W(t), material height h(t), weight gain rate ΔW(t) per unit time, and tilt angle θ(t) to form the cargo dataset D. L ;

[0059] The above data was subjected to time synchronization, outlier removal, missing value compensation, and normalization to obtain the standardized transportation dataset D. T ′ and standardized cargo dataset D L ′.

[0060] Horizontal continuous loaders, such as belt conveyors; downstream loading equipment, such as truck beds, hoppers, or bins.

[0061] The specific process of feature extraction based on standardized transportation datasets and standardized cargo datasets includes:

[0062] For the standardized transportation dataset D T A sliding time window is established with a fixed time length L, and the k-th time window is denoted as W. k Its starting time is t k The end time is t k +L, the window slides forward sequentially with a fixed step size, dividing the continuous time series into multiple analysis intervals; within each time window, statistical calculations are performed on the conveyor belt speed v(t), and the arithmetic mean of the speed values ​​of all sampling points within the window is calculated to obtain the average conveyor speed. This represents the overall conveying capacity level within that time interval; similarly, the instantaneous material flow rate q(t) is processed in the same way to obtain the average material flow rate. This represents the average scale of material transported per unit time within that time window.

[0063] After obtaining the average value, the standard deviation σ of the flow data is calculated to characterize the degree of fluctuation. q，k The specific calculation method is as follows:

[0064] Compare each sampled flow rate value within the window with the average flow rate. The sum of squared differences, divided by the number of sampling points within the window and then squared, yields a result reflecting the dispersion of the flow rate. The standard deviation σ is calculated similarly for the velocity data. v，k , used to reflect the intensity of velocity fluctuations;

[0065] A system stability index S is established based on flow fluctuations and velocity fluctuations. stab，k The system stability index is defined as the sum of the ratio of the standard deviation of flow rate to the average flow rate and the ratio of the standard deviation of speed to the average speed. The system stability index is used to measure the stability of the transportation end within a time window. When the value increases, it indicates that the system fluctuations are enhanced.

[0066] Furthermore, the difference between the maximum and minimum values ​​of the drive motor current I(t) within the time window is used to obtain the current variation amplitude I. var，k The magnitude of the current change characterizes the instantaneous load impact intensity; the average value T of the transmission tension T(t) is calculated within a certain window. avg，k The average value of the window reflects the overall stress level of the conveyor belt.

[0067] For the standardized cargo dataset D LPerform the same time window segmentation; calculate the weight gain rate Δ per unit time for the load-bearing weight W(t) of the loading equipment. Wrate，k The calculation process is as follows: subtract the weight at the beginning of the window from the weight at the end of the window, and then divide by the time length L. This parameter represents the actual rate at which the material enters the loading equipment during this time interval. The same difference method is used to calculate the rate of change of the material level h(t). rate，k The rate of change in material level characterizes the filling process of the loading space; the average value within the window of the loading attitude tilt angle θ(t) is calculated, and the average value of the absolute deviation of each sampling point relative to this average value is obtained to obtain the attitude offset θ. dev，k This attitude offset is used to measure the degree of off-center loading.

[0068] Based on the above, to enhance the ability to identify off-center load risks, the off-center load coupling factor K is calculated. couple，k It is defined as the attitude offset θ dev，k Weight gain rate per unit time ΔW rate，k The product of these factors will increase when the weight growth rate is high and the attitude deviates, thus reflecting potential off-center loading trends in advance. The weight at the end of the current window will be compared with the rated maximum load capacity W of the loading equipment. max The ratio is calculated to obtain the loading space utilization rate η. k Load space utilization rate indicates the current loading saturation level.

[0069] Finally, a transportation operation feature vector F is established. T，k ={ , , σ q，k S stab，k T avg，k I var，k and cargo status feature vector F L，k ={ΔW rate，k h rate，k θ dev，k K couple，k η k}, where the subscript k represents the corresponding sliding time window number.

[0070] The process of performing correlation analysis on horizontal continuous loaders and loading equipment based on transportation operation feature vectors includes:

[0071] S301: Obtaining the transportation operation feature vector F T，k Next, the material linear density function is established: the time window W is... k =[t k , t k+L Discretize the continuous data within the range into sampling points t. i Calculate the linear density ρ(t) of the material at each time step.i ): Through formula We obtain, where ρ(t) i q(t) represents the mass of material per unit length of conveyor belt. i v(t) represents the material flow rate at that moment. i () represents the conveyor belt speed.

[0072] S302: In time window W k Within, the continuous conveying process is divided into N k Given a time interval Δt, calculate the material mass m of each micro-unit. i，k That is: through formula m i，k =q(t i The result is obtained by Δt; then the spatial position x of each micro-unit is calculated. i，k That is: through the formula x i，k =v(t i )·(t i -t k The micro-unit sequence m is obtained and formed. k ,Right now: , where m i，k Indicates the mass of a micro-unit, reflecting the cumulative amount of material over a micro-time period; x i，k This indicates the position of the micro-unit on the conveyor belt.

[0073] S303: Based on micro-unit sequence m k Calculate the spatial gradient ▽ρ between adjacent micro-units. i，k Through the formula ▽ρ i，k =(ρ i+1，k -ρ i，k ) / (x i+1，k -x i，k ) is obtained; according to traffic flow theory, if ρ i，k >ρ c When this happens, a local congestion zone is formed; where ρ c The critical density is obtained through a density-flow rate fitting model. The relationship between flow rate q and density ρ is measured under different linear density conditions, and the fitting result is: q = aρ - bρ². When dq / dρ = 0, the corresponding density is the critical density ρc.

[0074] Next, calculate the propagation speed w of the congestion wave. i，k : Through formula w i，k =(q i+1 ,k-q i，k ) / (ρ i+1 ,k-ρ i，k ) obtained; w i，k This indicates the speed at which the accumulation wave propagates upstream or downstream;

[0075] S304: Congestion wave velocity w calculated based on S303 i，k Spatial gradient ▽ρ between adjacent micro-units i，k Predict the propagation distance x of the accumulated wave within a future time ΔT. cong，i，k : Through formula x cong，i，k =w i，k ·ΔT is obtained to determine the inlet distance D of the loading equipment. load Specifically, this comes from the equipment design parameters.

[0076] If the propagation distance of the accumulation wave is x cong，i，k ≥D load The accumulation wave will then reach the loading equipment and establish a flow mitigation factor β. i，k : Through formula β i，k =1-(ρ i，k -ρ c ) / (ρ max -ρ c ) is obtained, where ρ max The maximum mass that a conveyor belt can carry per unit length is calculated using the formula: ρ max =B·H0·ρ bulk Where: B is the effective width of the conveyor belt; H0 is the maximum stacking height; ρ bulk The material bulk density is then adjusted; subsequently, the micro-unit release time t is adjusted. i，k new : Through formula t i，k new =t i +β i，k ·Δt is obtained.

[0077] S305: The adjusted micro-unit mass and release time are combined into a matrix and used as the first result of joint control. .

[0078] The inversion correlation analysis based on the loading equipment and the horizontal continuous loader according to the cargo state feature vector includes the following process:

[0079] Obtain the cargo loading status feature vector F L，k Subsequently, this step, through three consecutive conversion processing stages, inverts the material distribution information at the loading end into an optimized strategy for the release of micro-units at the transportation end, thereby generating joint control result two and realizing bidirectional closed-loop regulation between the transportation end and the loading end.

[0080] S401: Obtain the cargo loading status feature vector F L，k Then, the loading space of the loading equipment is divided into a two-dimensional grid (x j y j The mass m of material per grid is calculated using the stacking height and unit volume density. j，k That is: through formula mj，k =f(h(x j y j ,t),ρ p ), where ρ p h(x, y, t) represents the particle density of the material; h(x, y, t) represents the stacking height of the ultrasonic ranging sensor. j y j ,t) represents the stacking height in the two-dimensional grid; based on the material mass of each grid, the material concentration field at the loading end is calculated using formula C. k ={C j，k}={m j，k / V j The material concentration field C at the carrier end is obtained. k V j This represents the mesh volume.

[0081] S402: Based on the material concentration field C at the loading end k And obtain the material diffusion coefficient D and the safe limit density ρ of the silo. load，max With critical density ρ c Calculate the diffusion potential energy and accumulation risk;

[0082] The material diffusion coefficient D is used to characterize the lateral migration ability of materials within the loading space. It is calculated as: D = k1·dp²·(1-ω)·cosφ, where: k1 is the empirical calibration coefficient; dp is the average particle size; ω is the moisture content; and φ is the internal friction angle. For highly adhesive materials, a correction term is added: D = D·e^(-k2·μa), where μa is the adhesion coefficient, k2 is the adhesion correction coefficient, e is the natural constant, and · is the multiplication sign.

[0083] For concentration field C k The Laplace operator is calculated using the formula Φ. j，k =D·▽ 2 C j，k Get Φ j，k Mapping the diffusion potential energy to a local stacking risk function : ;wherein Φ j，k Describes the local material concentration change trend, reflecting the diffusion tendency; Indicates the risk of localized accumulation, when When the value is greater than 1, it indicates that the grid may cause an impact on the transportation end;

[0084] S403: Obtain the micro-unit sequence Local accumulation risk function And the micro-unit mapping function g(i,j), where g(i,j) represents the loading end grid j corresponding to micro-unit i;

[0085] Next, based on the micro-cell mapping function g(i,j), the stacking risk of mapping each micro-cell i to the corresponding mesh j is determined. And adjust the release time The process of adjusting the release time is as follows: using the formula We obtain, where α is the adjustment coefficient; Δt0 is the fundamental time interval of the micro-unit. The delay is a time correction amount; t i This represents the release time of the i-th material micro-unit as it leaves the conveyor belt and enters the loading equipment space at the tail end of the horizontal continuous loader.

[0086] Finally, the adjusted micro-units are used to form a two-matrix joint control result. : .

[0087] The specific process for analyzing joint control result one and joint control result two is as follows:

[0088] After obtaining joint control result one and joint control result two, joint control result one is denoted as U. 1 The result of the joint control is denoted as U. 2 ; where: U 1 with U 2 All are control parameter vectors, with a unified structure defined as: U = ⟨v, q, θ>, where: v represents the belt speed of the horizontal continuous loader; q represents the instantaneous material conveying flow rate; and θ represents the loading equipment attitude or material placement angle adjustment parameter.

[0089] S501: For U 1 and U 2 Perform phase calibration; the specific phase calibration process is as follows:

[0090] Let t1 be the current control calculation time; Δt1 be the equivalent time delay caused by the material hysteresis characteristics of the loading equipment; U 2 (t1-Δt1) is the second result of the joint control after time correction;

[0091] Wherein, Δt1 is calculated from the average settling time of the material in the cargo state characteristic vector, that is, Δt1 is the average time from when the material enters the loading equipment to when it forms a stable stacking structure. After time mapping, the phase-aligned control pair is obtained: U 1 (t1) and U 2 (t1-Δt1);

[0092] S502: For U 1 with U 2 Differences exist in the control direction; entropy weight allocation is handled as follows:

[0093] In one instance, for example: U1 Increase speed; U 2 The speed needs to be reduced; therefore, a collision degree calculation is required.

[0094] Define the control difference vector: ΔU = U 1 -U 2 Its meaning is: the difference between two control results for each control parameter.

[0095] In this example, to avoid system oscillations caused by simple weighting, an uncertainty entropy value H is set to control the process.

[0096] H is calculated as follows: H = −Σp i1 ln(p i1 ); where: p i1 Let be the probability distribution value of the control parameter i1 within the historical stable operating range; i1 corresponds to the three control dimensions v, q, and θ, respectively. In this example, H represents the fluctuation complexity of the current control environment. If H is large, it indicates that the system is unstable, and the inversion control result U should be reduced. 2 The weight of H; if H is small, it indicates that the system is stable, then increase U. 2 Corrected weights.

[0097] Finally, the weighting coefficient α1 is defined as follows: α1 = 1 / (1+H), where the value of α1 ranges from 0 to 1; the larger H is, the smaller α1 is.

[0098] S503: Fusion calculation is performed using game theory optimization principles, specifically:

[0099] Let the final joint control result U* satisfy: U* = argminJ(U); where J(U) is the control cost function, defined as: J(U) = α1‖U-U¹‖² + (1-α1)‖U-U²‖² + β1R(U); where ‖·‖ represents the squared Euclidean distance; β1 is the system stability adjustment coefficient; and R(U) is the operational risk function. The operational risk function R(U) represents the impact of changes in control parameters on equipment load and safety, such as the rate of change of belt tension, the rate of change of hopper impact load, and other comprehensive risk indicators.

[0100] S504: In this example, to ensure that the control results conform to the actual equipment capability range, boundary correction is performed on U*, specifically: Let v min ≤v≤v max q min ≤q≤q max θ min ≤θ≤θ max If the calculation result exceeds the allowable range, it is projected to the nearest boundary value. Where v min v is the minimum permissible speed of the conveyor belt. max q is the maximum permissible speed of the conveyor belt.min q represents the minimum allowable instantaneous material conveying flow rate. max θ represents the maximum permissible instantaneous material conveying flow rate. min θ is the minimum permissible value of the loading equipment's attitude angle. max This represents the maximum permissible value for the attitude angle of the loading equipment.

[0101] The specific process of controlling the horizontal continuous loader and loading equipment based on the final joint control result is as follows:

[0102] S601: Based on the final joint control result U* obtained from S5, the time-series control vector U(t) is continuously mapped in the time dimension; U(t) = [U 12 (t), U 212 (t), ..., U n2 [(t)], where n2 is the number of control dimensions per U. i2 (t) represents the control output of the i2th execution control channel; U(t) is the joint control vector at time t.

[0103] It should be noted that U(t) is not a single control signal, but rather the result of time-continuous optimization control formed through multi-stage optimization from S2 to S5. Its generation process can be defined as: U(t) = Φ(X(t), , W(t), Θ(t)), where: X(t) is the real-time acquired state vector, For the twin prediction state, W(t) is the dynamic weight matrix, and Θ(t) is the control constraint parameter set and the joint optimization mapping function of Φ(·).

[0104] Expanding the expression, it can be defined as: U(t) = W(t)·u(t) + ΔU(t), where: u(t) = [u 12 (t), u 22 (t), …, u n2 [(t)] represents the independent control variables of each subsystem; It is a dynamic weight matrix that is updated in real time with error feedback;

[0105] ΔU(t)=K e ·e(t)+K c • C(t) is the correction compensation term, where: e(t) is the twin error vector, C(t) is the causal correlation matrix compression result, and K e K c All are adjustment coefficients.

[0106] S602: Based on the physical response state of the loading equipment under the action of U(t), the operating parameters of the equipment are collected in real time to form a response vector y(t): y(t) = [y 12 (t), y 22(t), ..., y m2 [(t)], and time-align the control input and response output to obtain the control-to-response mapping sequence Z(t): Z(t) = {U(t), y(t)}. Where: y(t) ∈ R m2 m2 is the number of dimensions of the state variables; y i2 (t) is the measurement value of the i2th state variable at time t.

[0107] S603: Based on the control-to-response mapping sequence Z(t), re-execute steps S2 to S5 to obtain the updated twin prediction output. : = And calculate the prediction error e(t): e(t) = y(t) - Wherein, the error is the weighted state-space error: e(t) = (y(t) - ) T Q(y(t)- ), where Q is the dynamic sensitivity matrix.

[0108] S604: Based on the error time series e(t), establish the error evolution gradient ▽e(t): ▽e(t)=e(t)-e(t-Δt);

[0109] Based on the error value and the rate of change of error, an anomaly discrimination function Ψ(t) is established: Ψ(t) = αe(t) + β▽e(t). When Ψ(t) > Θ0(t), time t is numbered as the k1th anomaly candidate point; that is, t = t k1 Therefore: e k1 =e(t k1 ); where the threshold Θ0(t) is dynamically generated based on the historical stable operating interval.

[0110] S605: Based on the set of anomaly candidate points, establish the control response coupling perturbation matrix. : , where Δt2 is the time delay from control to response.

[0111] Calculate the anomaly source contribution vector S based on the perturbation matrix. k1 : Through formula We obtain and extract the set of dominant disturbance sources Ω: Ω = {k1 | S} k1 >μ}. Among them, To control the response coupling perturbation matrix When the k-th state variable is determined to be the dominant anomaly variable, the k1-th row of the matrix is ​​taken to obtain... .

[0112] S606: Based on the abnormal error value e k1Disturbance contribution S k1 and joint control offset ΔU k1 ΔU k1 =U k1 (t)-U k1 (t-Δt2); Establish a comprehensive ranking function, using formula L k1 =λ1×e k1 +λ2×S k1 +λ3×∣ΔU k1 | Obtain the comprehensive ranking function L k1 ; where λ1, λ2, and λ3 are all preset weighting coefficients;

[0113] According to L k1 Size classification: When L k1 >L s If L is found to be severely abnormal, it is considered a serious abnormality. m <L k1 ≤L s If it is, it is judged as a moderate abnormality, when L l <L k1 ≤L m If so, it is judged as a minor abnormality and when L k1 ≤L l If it is, then it is considered normal; among them, L l L is the threshold for minor anomalies. m For medium anomaly threshold, L s This is the threshold for severe anomalies.

[0114] S607: Establishing a processing cost function J based on anomaly level results k1 That is: J k1 =C risk (L k1 )+C adjust (U k1 )+C delay , where C risk (L k1 The risk cost function represents the potential system risk and loss if this anomaly is not addressed; it is defined as a monotonically increasing function: C risk (L k1 )=α1×L k1 2 Where: α3 > 0, which is the risk amplification factor; the higher the risk level, the greater the risk cost; C adjust (U k1 The adjustment cost function represents the energy, mechanical wear, or labor costs consumed in performing equipment adjustments, and is defined as: C adjust (U k1 )=β∥U k1 -U ref ∥ 2, where: U ref The preset reference stable control vector, ||·|| Euclidean norm, β3>0, is the adjustment cost coefficient; C delay The delay cost represents the time loss cost incurred during processing. It is defined as: C delay =γ3×Δt k1 Where: γ3>0, is the unit time delay cost coefficient, Δt k1 The time required to handle the k1th exception.

[0115] When J k1 >J limit If the anomaly is detected, the abnormality will be incorporated into the joint control recalculation process, triggering the corresponding loading facility adjustment action; otherwise, the twin model parameters will be corrected, and physical facility actions will not be executed. Wherein J... limit The maximum processing cost allowed by the preset system.

[0116] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for slow loading and fast unloading of bulk materials, characterized in that, include: S1: Real-time data collection of the operation of the horizontal continuous loader and its downstream loading equipment is carried out through multi-source sensors to establish a transportation dataset and a cargo dataset. The collected raw data is then cleaned and standardized to obtain a standardized transportation dataset and a standardized cargo dataset. S2: Based on the standardized transportation dataset and the standardized cargo dataset, extract the transportation operation feature vector and the cargo status feature vector respectively; S3: Based on the characteristic vector of transportation operation, traffic flow theory and congestion wave propagation model are set up to predict and regulate the flow state of materials on the horizontal continuous loader and generate joint control result one. S4: Based on the cargo loading state feature vector, the material distribution inside the loading equipment is inverted through material concentration field modeling and diffusion potential energy analysis to generate joint control result two; S5: Perform phase alignment, weight allocation, and game-theoretic fusion optimization on joint control result one and joint control result two to obtain the final joint control result; S6: Based on the final joint control results, perform closed-loop control on the horizontal continuous loader and loading equipment, monitor the control response in real time, identify and trace anomalies, and dynamically adjust the control strategy.

2. The method for slow loading and fast unloading of bulk materials according to claim 1, characterized in that, The specific process for extracting the transportation operation feature vector and the cargo status feature vector respectively is as follows: By establishing sliding time windows with fixed time lengths for standardized transportation datasets and standardized cargo datasets, and sliding each window forward with a fixed step size, the continuous time series is divided into multiple analysis intervals. Within each time window, statistical calculations are performed on the conveyor belt speed, instantaneous material flow rate, drive motor current, and conveying tension in the transportation dataset to obtain the transportation operation feature vectors. Within the same time window, differential and statistical operations are performed on the load weight, material height, and loading posture tilt angle in the cargo dataset to obtain the cargo state feature vectors.

3. The method for slow loading and fast unloading of bulk materials according to claim 2, characterized in that, The specific process of generating joint control result one includes: The average conveying speed and average material flow rate are obtained from the transportation operation feature vector, and a material linear density function is established. The continuous conveying process within the time window is divided into multiple micro-time periods. The mass of each micro-unit and its spatial position on the conveyor belt are calculated to form a micro-unit sequence. Based on the micro-unit sequence, the spatial gradient of adjacent micro-units is calculated, and local congestion areas are identified according to traffic flow theory to calculate the congestion wave propagation speed. Based on the congestion wave propagation speed and spatial gradient, the propagation distance of the accumulation wave in the future time is predicted; if the accumulation wave propagation distance reaches or exceeds the inlet distance of the loading equipment, the flow rate mitigation factor is calculated, and the micro-unit release time is adjusted accordingly; the adjusted micro-unit mass and release time are combined to form the joint control result one.

4. The method for slow loading and fast unloading of bulk materials according to claim 3, characterized in that, The specific process of generating joint control result two includes: Material level height data is obtained from the feature vector of the loading status. The space of the loading equipment is divided into a two-dimensional grid. The material mass of each grid is calculated based on the stacking height and unit volume density, and a material concentration field at the loading end is established. The material diffusion coefficient is calculated based on the material characteristic parameters, and the Laplace operator is calculated for the material concentration field at the loading end to obtain the diffusion potential energy. The diffusion potential energy is then mapped to the local stacking risk function. A mapping function between micro-units and loading end meshes is established to map each micro-unit in the micro-unit sequence to the local stacking risk value of the corresponding mesh, and the micro-unit release time is adjusted accordingly to form joint control result two.

5. The method for slow loading and fast unloading of bulk materials according to claim 4, characterized in that, The specific process for obtaining the final joint control result includes: Phase calibration is performed on joint control result one and joint control result two. Time delay correction is performed on the second joint control result based on the average material settling time in the cargo state characteristic vector to obtain a time-aligned control pair. Calculate the control difference vector and obtain the weight coefficients of the second joint control result based on the control entropy value. The more unstable the system, the greater the weight of the first joint control result. Establish a control cost function, and obtain the final joint control result by minimizing the control cost function; perform boundary correction on the final joint control result.

6. The method for slow loading and fast unloading of bulk materials according to claim 5, characterized in that, The process of performing closed-loop control on the horizontal continuous loader and loading equipment based on the final joint control results is as follows: Based on the final joint control results, a timing control vector is generated, and the device response vector is collected in real time to form a control response mapping sequence; Based on the control response mapping sequence, steps S2 to S5 are re-executed to obtain the twin prediction output and calculate the prediction error; An error evolution gradient is established based on the prediction error time series, and an anomaly discrimination function is established. When the anomaly detection function value exceeds the dynamic threshold, it is identified as an anomaly candidate point, and the error value at the corresponding time is recorded; Based on the anomaly candidate points, a control response coupling perturbation matrix is ​​established, the contribution of anomaly sources is calculated, and the set of dominant perturbation sources is extracted. Based on the abnormal error value, the contribution of the abnormal source, and the control offset, a comprehensive grading function is established to classify the abnormality level; Establish a processing cost function. If the processing cost exceeds the preset maximum allowable cost, the exception will be incorporated into the joint control to recalculate the process and trigger the corresponding loading facility adjustment action. Otherwise, the twin model parameters will be corrected.

Images