Intelligent production scheduling method and system for noble metal processing

Abstract

The present application relates to the technical field of scheduling optimization, in particular to a noble metal processing intelligent production scheduling method and system, comprising the following steps: collecting the geometric characteristics of noble metal blanks, extracting numerical calculation quality specific surface area, fitting the oxidation kinetics model to calculate the surface reaction evolution rate, accumulating time sequence loss to screen the minimum value item to determine the low loss processing path, calculating the motor torque and the surface reduction rate to obtain the heat surplus state quantity, comparing the temperature threshold value and embedding the shutdown cooling trigger pulse at the process time node. In the present application, the surface oxidation loss value in the multi-channel processing service process is quantitatively analyzed, the optimal search is used to lock the minimum value processing path, the oxidation and volatilization loss of noble metal materials in the high temperature environment is reduced, the dynamic heat surplus state of the real-time feature parameter sensing equipment is fused, the risk of internal microstructure defects caused by excessive accumulation of deformation heat is eliminated, and the execution efficiency and controlled accuracy of the processing service scheduling control sequence are improved.

Description

Technical Field

[0001] This invention relates to the field of scheduling optimization technology, and in particular to an intelligent production scheduling method and system for precious metal processing. Background Technology

[0002] The field of scheduling optimization technology mainly studies the theories and methods for coordinating task sequence, operation time, and resource allocation under given resource constraints. Core aspects include the formulation of operation sequencing rules, equipment load balancing, material supply cycle matching, constraints on process connection relationships, multi-resource collaborative scheduling, and production cycle compression. Its technical system involves establishing a production model that includes parameters such as process processing time, equipment availability time, personnel shift arrangement, and material arrival time. A complete production scheduling scheme is formed by determining process priority, constructing a time sequence table, calculating equipment occupancy intervals, and adjusting task insertion positions. It is widely used in discrete manufacturing, process manufacturing, and multi-variety, small-batch production scenarios.

[0003] The traditional intelligent production scheduling method for precious metal processing refers to the method of production planning for the processing of precious metals such as gold, silver, and platinum through multiple processes such as smelting, rolling, wire drawing, stamping, and polishing. The technical issues it addresses are determining the processing sequence and time nodes of each batch of workpieces under conditions of fluctuating order quantities, diverse processing routes, strong equipment specialization, and high raw material value. The traditional method breaks down customer orders into several production batches based on manual planning tables, and records them one by one in paper or electronic spreadsheets according to the process flow. Then, the start and end times are manually calculated based on the daily processing time of the equipment, and the raw material balance is checked in conjunction with inventory records. If equipment conflicts occur, the batch order is adjusted or some process times are delayed to rearrange the batches. At the same time, the process time is estimated based on historical processing records and used as the basis for scheduling, thereby forming a daily or weekly production schedule.

[0004] Traditional precious metal processing service scheduling relies on manual experience to allocate process time, estimate time based on historical records and manually adjust equipment occupancy. It ignores the impact of dynamic changes in the geometric shape of the billet on surface oxidation loss during processing, and cannot monitor the relationship between environmental parameters and material oxidation kinetics in real time. This results in uncontrollable oxidation loss of materials during multi-process flow, does not consider the impact of deformation heat accumulation from machining on the stability of metal microstructure, and lacks dynamic intervention methods for thermal equilibrium state. This causes the scheduling to be out of sync with the actual evolution process and reduces material utilization. Summary of the Invention

[0005] To address the technical problems of traditional precious metal processing service scheduling relying on manual experience to allocate process times, estimating time based on historical records and manually adjusting equipment occupancy, ignoring the impact of dynamic changes in the geometric shape of the billet on surface oxidation loss during processing, failing to monitor the correlation between environmental parameters and material oxidation kinetics in real time, resulting in uncontrollable oxidation loss of materials during multi-process flow, not considering the impact of deformation heat accumulation from machining on the stability of metal microstructure, and lacking dynamic intervention methods for thermal equilibrium state, causing a disconnect between scheduling and actual evolution process, and reducing material utilization, this invention provides an intelligent production scheduling method for precious metal processing.

[0006] To achieve the above objectives, this invention employs an intelligent production scheduling method for precious metal processing, comprising the following steps: S1: Collect the length, width and diameter of the precious metal billet, perform specific surface area mapping transformation based on the geometric shape characteristics of the precious metal billet, calculate the ratio between the transformation result and the mass of the precious metal billet, and generate the mass specific surface area. S2: Call the mass specific surface area, collect the furnace lining temperature and oxygen partial pressure inside the melting furnace, perform fitting calculation through the high-temperature oxidation kinetic model, calculate the oxidation reaction rate on the surface of the precious metal billet, and generate material oxidation rate characteristic value; S3: Call the material oxidation rate characteristic value, retrieve the exposure time from the preset process path library, perform time-series loss accumulation calculation, and select the path with the highest total loss from the preset process path library to determine the low-loss processing path. S4: For the low-loss processing path, collect the motor torque and material reduction rate and perform vector multiplication. Combine the temperature drop rate parameter to check the thermal balance and obtain the material thermal surplus state of the equipment. S5: Read the process time node of the low-loss processing path, perform a comparison and determination of the material thermal surplus state of the equipment and the recrystallization temperature threshold, embed a shutdown cooling trigger pulse at the process time node, and generate a production scheduling control sequence.

[0007] As a further aspect of the present invention, the specific surface area includes surface roughness, material porosity, and geometric shape factor; the material oxidation rate characteristic value includes oxidation weight gain, oxide film thickness, and parabolic rate constant; the low-loss processing path includes process flow sequence, processing lead time, and station dwell time; the equipment material thermal surplus state quantity includes material deformation heat, interfacial frictional heat, and system heat loss; and the production scheduling control sequence includes production cycle time, start-up time, and shutdown cooling time.

[0008] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Collect a set of three-dimensional coordinate point clouds of metal billet, filter outliers from the three-dimensional coordinate point cloud set, extract the outer edge contour points, calculate the extreme difference of the outer edge contour points in the coordinate axis projection direction, obtain the entity length, width and diameter values, and combine them to establish a spatial three-dimensional feature matrix. S102: Call the spatial three-dimensional feature matrix, calculate the entity length value, entity width value and entity diameter value based on the integral operator to generate the outer surface area value, calculate the internal volume space value based on the volume operator, divide the outer surface area value by the internal volume space value to obtain the surface area ratio mapping value. S103: The monitoring base sensor acquires the weight data of the metal billet, performs a division operation on the surface area ratio mapping value and the weight data of the metal billet to obtain a ratio operation number sequence, performs bit truncation and recombination on the ratio operation number sequence according to the preset precision verification floating-point base comparison value, and generates mass specific surface area.

[0009] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Call the mass specific surface area, collect the furnace lining temperature distribution sequence and oxygen partial pressure value, perform normalization mapping on the mass specific surface area, furnace lining temperature distribution sequence and oxygen partial pressure value to obtain environmental feature vector, and establish furnace oxidation potential energy matrix in combination with preset activation energy constant. S202: Extract the row vectors of the furnace oxidation potential energy matrix to construct a multidimensional potential energy space, perform a product operation on the mass specific surface area and the furnace oxidation potential energy matrix to obtain the reaction intensity vector, extract the sliding window discrete change amount from the reaction intensity vector, and generate a basic reaction evolution rate sequence. S203: Call the basic reaction evolution rate sequence, perform continuous integration operation on the basic reaction evolution rate sequence and the preset time decay factor to extract the multidimensional curve extreme value set, perform arithmetic mean calculation on the multidimensional curve extreme value set to extract the global smoothing index, and generate material oxidation rate characteristic value.

[0010] As a further aspect of the present invention, the step of extracting the discrete change amount of the sliding window for the reaction intensity vector refers to setting the sampling length and displacement step of the sliding window, truncating the reaction intensity vector according to the sampling length to obtain a vector subsequence, extracting the difference between the last value and the first value of the vector subsequence, dividing the difference by the sampling length to obtain the local rate of change, translating the sliding window according to the displacement step, and collecting multiple sets of local rate of change combinations to generate the discrete change amount of the sliding window.

[0011] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Call the material oxidation rate feature value, read the discrete stage exposure time sequence in the process path library, perform element-wise multiplication based on the material oxidation rate feature value and the discrete stage exposure time sequence to obtain the node loss feature value set, and perform topological mapping according to the node number to establish a single-segment oxidation loss matrix; S302: Call the single-segment oxidation loss matrix, read the local preset initial oxidation increment offset, perform time-series integral accumulation operation on the row vector to obtain the path process loss distribution vector, perform arithmetic summation on the path process loss distribution vector and the initial oxidation increment offset to establish the candidate path loss total. S303: Call the total loss of the candidate path to perform bubble sort, obtain the linked list of loss values ​​in ascending order, retrieve the first and smallest value item from it, associate the identification code with the smallest value item, and generate a low-loss processing path.

[0012] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: For the low-loss processing path, collect the torque continuous state sequence and the set of surface reduction rate feature parameters, and perform time-series alignment to extract feature samples of the torque continuous state sequence and the set of surface reduction rate feature parameters according to the timestamp of the low-loss processing path node to establish a dynamic mapping matrix. S402: Call the dynamic mapping matrix, extract the torque vector and deformation vector inside the dynamic mapping matrix, perform parametric fusion analysis on the torque vector and deformation vector to obtain the instantaneous work equivalent sequence, and perform smoothing and noise reduction on the instantaneous work equivalent sequence by truncating a segment along the time axis to obtain the steady-state heat load vector. S403: Based on the steady-state heat load vector, collect the temperature drop rate parameter and read the specific heat capacity constant. Construct a heat dissipation benchmark threshold based on the thermal correlation characteristics between the temperature drop rate parameter and the specific heat capacity constant. Compare the steady-state heat load vector with the heat dissipation benchmark threshold to extract the deviation set and generate the equipment material heat surplus state quantity.

[0013] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Call the low-loss processing path to extract the process time node distribution vector, call the equipment material heat surplus state quantity and preset recrystallization temperature threshold to perform over-limit state analysis to generate heat overflow feature item, associate time axis coordinates to assign high-level logic placeholders, and establish over-limit state marker vector. S502: Call the over-limit state marker vector, read the preset response delay period constant and perform timing forward parsing on the coordinate position of the high-level logic placeholder to extract the early trigger timestamp, configure the signal amplitude and signal pulse width for the early trigger timestamp and perform waveform fitting to generate a shutdown cooling trigger pulse; S503: Invoke the shutdown cooling trigger pulse, obtain the basic execution timing list, embed the shutdown cooling trigger pulse into the idle time slot of the basic execution timing list according to the advance trigger timestamp clock cycle, obtain the hybrid scheduling instruction signal set, perform anti-collision verification to eliminate overlapping time windows, and obtain the production scheduling control sequence of precious metal processing services.

[0014] As a further aspect of the present invention, the preset recrystallization temperature threshold is determined by parsing the attribute identifier of the material to be processed and performing addressing comparison in a preset material thermophysical constant mapping table to extract the critical temperature value of the target material.

[0015] A smart production scheduling system for precious metal processing includes: The geometric feature mapping module collects the length, width, and diameter of the precious metal billet, performs a specific surface area mapping transformation on the geometric shape features of the precious metal billet, calculates the ratio between the transformation result and the mass of the precious metal billet, and generates the mass specific surface area. The oxidation rate fitting module calls the mass specific surface area, collects the furnace lining temperature and oxygen partial pressure inside the melting furnace, performs fitting calculations through a high-temperature oxidation kinetic model, calculates the oxidation reaction rate on the surface of the precious metal billet, and generates material oxidation rate characteristic values. The loss timing optimization module calls the material oxidation rate characteristic value, retrieves the exposure time from the preset process path library, performs timing loss accumulation calculation, and selects the path with the highest total loss from the preset process path library to determine the low loss processing path. The thermal balance state verification module collects motor torque and material reduction rate for the low-loss processing path and performs vector multiplication calculation. It then verifies the thermal balance by combining the temperature drop rate parameter and obtains the material thermal surplus state of the equipment. The scheduling pulse generation module reads the process time nodes of the target low-loss processing path, performs a comparison and determination of the material thermal surplus state of the equipment and the recrystallization temperature threshold, embeds a shutdown cooling trigger pulse at the process time node, and generates a production scheduling control sequence.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the specific surface area of ​​precious metal billets is obtained by mapping and transforming their geometric features. An oxidation kinetic model is constructed by combining furnace environmental parameters to calculate the surface reaction rate, thereby achieving quantitative analysis of oxidation loss in different process paths. By accumulating time-series losses and filtering for minimum values, low-loss processing paths are locked, reducing the volatilization and oxidation loss of precious metals in high-temperature processes. Dynamic thermal balance verification is performed by integrating motor torque and material surface area reduction rate characteristics to obtain the real-time thermal surplus state of the material. The recrystallization temperature threshold is determined by comparing material properties. A shutdown cooling trigger pulse is adaptively embedded at process time nodes to eliminate the structural damage caused by excessive accumulation of deformation heat during processing. This achieves deep coupling between scheduling instructions and material physical properties, improving control accuracy. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the accompanying drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0020] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0021] Please see Figure 1 This invention provides an intelligent production scheduling method for precious metal processing, comprising the following steps: S1: Collect the length, width and diameter of the precious metal billet, perform specific surface area mapping transformation based on the geometric shape characteristics of the precious metal billet, calculate the ratio between the transformation result and the mass of the precious metal billet, and generate the mass specific surface area. S2: Call the mass specific surface area, collect the furnace lining temperature and oxygen partial pressure inside the melting furnace, perform fitting calculation through the high-temperature oxidation kinetic model, calculate the oxidation reaction rate on the surface of the precious metal billet, and generate material oxidation rate characteristic values. S3: Call the material oxidation rate characteristic value, retrieve the exposure time from the preset process path library, perform time-series loss accumulation calculation, and select the path with the highest total loss from the preset process path library to determine the low-loss processing path. S4: For low-loss processing paths, collect motor torque and material reduction rate and perform vector multiplication calculation, combine with temperature drop rate parameter to check thermal balance, and obtain the material thermal surplus state of the equipment. S5: Read the process time node of the low-loss processing path, perform a comparison and judgment between the equipment material thermal surplus state and the recrystallization temperature threshold, embed a shutdown cooling trigger pulse at the process time node, and generate a production scheduling control sequence.

[0022] Mass specific surface area includes surface roughness, material porosity, and geometric shape factor; material oxidation rate characteristic values ​​include oxidation weight gain, oxide film thickness, and parabolic rate constant; low-loss processing path includes process flow sequence, processing lead time, and station dwell time; equipment material thermal surplus state quantity includes material deformation heat, interfacial frictional heat, and system heat loss; production scheduling control sequence includes production cycle time, start-up time, and shutdown cooling time.

[0023] Please see Figure 2 The specific steps of S1 are as follows: S101: Collect a set of three-dimensional coordinate point clouds of metal billet, filter outliers from the three-dimensional coordinate point cloud set, extract the outer edge contour points, calculate the extreme difference of the outer edge contour points in the coordinate axis projection direction, obtain the entity length, width and diameter values, and combine them to establish a spatial three-dimensional feature matrix. A 3D high-frequency laser scanner located on top of the processing table is directly invoked to perform an all-around optical scan of the stationary gold billet at a scanning frequency of 1200 lines per second. This acquires 500,000 discrete spatial point data points with horizontal, vertical, and triangular coordinates, which are then aggregated to construct an initial 3D coordinate point cloud set for the metal billet. For this 3D coordinate point cloud set, spatial filtering logic based on a distance threshold is invoked. The neighborhood determination radius is set to 5 mm, and the core point cloud quantity threshold is set to 15. The number of neighboring points within the neighborhood determination radius for each coordinate point is calculated. Spatial coordinate points with fewer than 15 neighboring points are directly removed from the set, thus filtering out outliers caused by environmental dust or light refraction. After filtering outliers, for the remaining valid point cloud set, a spatial bounding box traversal operation is invoked. All points are traversed along the three orthogonal directions of the Cartesian coordinate system, extracting the maximum and minimum points of the horizontal, vertical, and triangular coordinates, and assigning them to the outer edge contour point set. After extracting the outer edge contour points, their extreme points in each coordinate axis projection direction are obtained, and arithmetic subtraction is performed to calculate the extreme value difference. For example, the maximum horizontal coordinate of the outer edge contour points in the horizontal axis direction is 250 mm and the minimum horizontal coordinate is 50 mm. Subtracting the maximum and minimum horizontal coordinates gives the entity length of 200 mm. The maximum vertical coordinate of the outer edge contour points in the vertical axis direction is 120 mm and the minimum vertical coordinate is 40 mm. Subtracting the maximum and minimum vertical coordinates gives the entity width of 80 mm. The maximum vertical coordinate of the outer edge contour points in the vertical axis direction is 90 mm and the minimum vertical coordinate is 30 mm. Subtracting the maximum and minimum vertical coordinates gives the entity diameter of 60 mm. After acquiring the above data, the entity length (200 mm), width (80 mm), and diameter (60 mm) are arranged in row-major order and stored in a single-column 3D data array, thereby creating a spatial 3D feature matrix characterizing the geometric properties of the billet. The advantage of this computational logic is that by directly extracting the coordinate projection extrema and performing difference calculations, the computational power required for full-dimensional surface fitting is significantly reduced, improving the response speed of preliminary processing services and enabling the processing service center to complete the calibration of the material profile within a shorter lifecycle.

[0024] S102: Call the three-dimensional feature matrix of space, calculate the entity length, entity width and entity diameter values ​​based on the integral operator to generate the external surface area value, calculate the internal volume space value based on the volume operator, divide the external surface area value by the internal volume space value to obtain the surface area ratio mapping value. The geometric parameters stored within the constructed 3D spatial feature matrix are read. For the entity's length, width, and diameter, the edge length of the outer surface curvature micro-element is preset to 0.1 mm. A 2D micro-element mesh is constructed along the entity's surface contour in a Cartesian coordinate system. The outer surface area integral operator is used to continuously accumulate the local area values ​​of all 2D micro-element meshes to generate the outer surface area value. Subsequently, the entity's internal 3D space is divided into 3D voxel units with an edge length of 0.1 mm. The internal volume space integral operator is used to perform a 3D summation calculation on the volume values ​​of all 3D voxel units to generate the internal volume space value. For example, the outer surface area of ​​the metal billet is calculated to be 48,000 square millimeters using the outer surface area integral operator, and the internal volume space is calculated to be 120,000 cubic millimeters using the internal volume space integral operator. After obtaining these two key parameters, the external surface area is used as the dividend, and the internal volume space is used as the divisor. An arithmetic division operation is performed to obtain the initial volumetric surface area ratio. This result is then multiplied by the material density baseline mapping coefficient of 0.8 to obtain the surface area ratio mapping value. Substituting the data from the previous example, dividing 48,000 square millimeters by 120,000 cubic millimeters yields an initial volumetric surface area ratio of 0.4. Multiplying 0.4 by the material density baseline mapping coefficient of 0.8 gives a surface area ratio mapping value of 0.32. The advantage of this calculation logic is that by introducing the material density baseline mapping coefficient to weight and correct the ratio of external surface area to internal volume, it can effectively shield the volume measurement error caused by the porosity of the material itself. This lays a solid physical calculation foundation for providing high-standard customized material processing services, ensuring that the multiple feature models output by the processing services have extremely high fidelity.

[0025] S103: The monitoring base sensor acquires the weight data of the metal billet, performs a division operation on the surface area ratio mapping value and the weight data of the metal billet to obtain the ratio operation digital sequence, performs bit truncation and recombination on the ratio operation digital sequence according to the preset precision verification floating-point base comparison value, and generates mass specific surface area; The system reads 100 real-time pressure conversion electrical signals collected by a gravity pressure sensor installed under the support base within a 5-second time window using an industrial communication protocol. After removing the highest and lowest values ​​from these 100 signals, an arithmetic average is calculated, and this average is combined with a preset sensor gravity conversion coefficient of 2.5 to obtain the current weight data of the metal billet. After obtaining the output surface area ratio mapping value, this value is used as the dividend, and the filtered metal billet weight data is used as the divisor to perform an arithmetic division operation to obtain a ratio sequence. For example, if the calculated surface area ratio mapping value is 0.32, and the weight data of the metal billet obtained and converted by the base sensor is 4 kg, the resulting ratio sequence after division is 0.08. To meet the instruction bit width limitations of high-precision manufacturing systems, a preset precision check floating-point base of 0.001 is read from the local storage medium. The significant digits after the decimal point of the previously obtained ratio operation sequence are extracted. The length of these significant digits is compared with the three decimal places corresponding to the preset precision check floating-point base. If the number of decimal places in the ratio operation sequence exceeds three, a forced truncation operation is performed starting from the fourth digit, discarding the trailing value. The remaining value is then reassembled to generate the final mass specific surface area used to characterize the material's unit mass exposure. In the above example, the ratio operation sequence 0.08 has two decimal places, which is less than three, so truncation is unnecessary, and 0.08 is directly recognized as the mass specific surface area. The advantage of this operation logic is that, through division mapping combined with a forced truncation and reassembly mechanism, it ensures the consistency of data length in subsequent input control buses, avoiding floating-point overflow. This improves the real-time data throughput capability of cloud-based CNC machining services, ensures zero-latency interaction between the underlying sensing nodes and the high-level machining service instruction scheduler, and thus establishes a reliable channel for seamless data communication between machining service nodes.

[0026] Please see Figure 3 The specific steps of S2 are as follows: S201: Call the specific surface area of ​​mass, collect the furnace lining temperature distribution sequence and oxygen partial pressure value, perform normalization mapping on the specific surface area of ​​mass, furnace lining temperature distribution sequence and oxygen partial pressure value to obtain environmental feature vector, and establish the furnace oxidation potential energy matrix in combination with the preset activation energy constant. The system retrieves the generated mass surface area value from the memory buffer area. Simultaneously, it acquires temperature values ​​from various observation points within the furnace at a frequency of twice per second using a thermocouple array embedded in the furnace wall, combining these values ​​to form a furnace lining temperature distribution sequence. It also uses a zirconia oxygen probe to obtain the current oxygen partial pressure value inside the furnace. After acquiring these three environmental characteristics, a unified standard interval is set with a lower limit of 0 and an upper limit of 1. For the mass surface area, the maximum temperature value in the furnace lining temperature distribution sequence, and the oxygen partial pressure value, the current value of each parameter is subtracted from the preset theoretical minimum value, and then divided by the difference between the preset theoretical maximum and theoretical minimum values. A normalization mapping operation is then performed to obtain three decimals within the range of 0 to 1, forming an environmental feature vector. For example, if the current oxygen partial pressure is 0.04 MPa, the preset theoretical maximum is 0.1 MPa and the minimum is 0, then the normalized result is 0.4. Similarly, the normalized value of the mass specific surface area is 0.8 and the normalized value of the maximum temperature is 0.6, thus obtaining an environmental feature vector containing an array of three elements: 0.8, 0.6, and 0.4. Next, a preset activation energy constant for a specific precious metal material is retrieved. This activation energy constant is determined experimentally by measuring the critical potential energy of the baseline reaction; here, it is read as 150 kJ / mol. Each element in the environmental feature vector is then multiplied arithmetically with this preset activation energy constant. The three product results are then arranged according to the element order of the environmental feature vector, with the diagonal elements padded with 0s, thus establishing a 3x3 furnace oxidation potential energy matrix. In a specific example, 0.8 multiplied by 150 yields 120, 0.6 multiplied by 150 yields 90, and 0.4 multiplied by 150 yields 60. The diagonal values ​​of the constructed furnace oxidation potential energy matrix are 120, 90, and 60, respectively. Constructing a high-dimensional potential energy matrix is ​​a core step in improving the overall response efficiency of the intelligent processing service platform, enabling various online processing services to solve for service status evaluation indicators based on a unified potential energy field.

[0027] Table 1: Environmental Characteristic Parameters and Normalization Results; As shown in Table 1, the environmental characteristic parameters are normalized and mapped to a dimensionless decimal range after passing through the set extreme value boundaries, providing standard data support for subsequent potential energy matrix calculations.

[0028] S202: Extract the row vectors of the furnace oxidation potential energy matrix to construct a multidimensional potential energy space. Perform a product operation on the mass specific surface area and the furnace oxidation potential energy matrix to obtain the reaction intensity vector. Extract the discrete change of the sliding window from the reaction intensity vector to generate the basic reaction evolution rate sequence. The generated furnace oxidation potential energy matrix is ​​read, and the first, second, and third row vectors of this matrix are extracted as three orthogonal bases of the spatial coordinate system to construct a multidimensional potential energy space. Then, the specific surface area obtained in the previous step is used as a scalar coefficient and numerically multiplied with the main diagonal elements of the furnace oxidation potential energy matrix to obtain a reaction intensity vector containing three product values. For example, a specific surface area of ​​0.08, multiplied by the main diagonal elements 120, 90, and 60 respectively, yields a reaction intensity vector containing the values ​​9.6, 7.2, and 4.8. Based on this, a sliding window discrete change is extracted from the continuous observation sequence of this reaction intensity vector over time. The sliding window is set to a sampling length of 10 time periods and a displacement step of 2 time periods. The continuous observation sequence of the reaction intensity vector is truncated according to the sampling length of 10 to obtain a vector subsequence containing 10 values. The last value in this subsequence is extracted and subtracted from the first value to obtain the difference. This difference is then divided by the sampling length of 10 to obtain the local rate of change within the window. For example, the first value of the extracted vector subsequence is 9.6 and the last value is 11.6, the difference is 2.0, and 2.0 divided by 10 gives a local rate of change of 0.2. The sliding window is then shifted by a displacement step of 2, and the second local rate of change is calculated again. This process is repeated until the entire sequence is traversed. Multiple sets of local rates of change are collected and combined in chronological order to generate the discrete change value of the sliding window. This discrete change value is directly output as the basic reaction evolution rate sequence. The advantage of this operation logic is that by using the difference between the first and last values ​​extracted by the sliding window and dividing it by the step size to calculate the local rate of change, interference caused by high-frequency local oscillations is effectively filtered out. This ensures that the feedback dynamic boundary of the underlying control unit is within a stable and safe range when it invokes processing services.

[0029] S203: Call the basic reaction evolution rate sequence, perform continuous integration operation on the basic reaction evolution rate sequence and the preset time decay factor to extract the multidimensional curve extreme value set, perform arithmetic mean calculation on the multidimensional curve extreme value set to extract the global smoothing index, and generate material oxidation rate characteristic value; The generated basic reaction evolution rate sequence is retrieved from the local register, and a preset time decay factor, set to 0.95 based on empirical values ​​of heat dissipation, is read simultaneously. Each discrete value in the basic reaction evolution rate sequence is multiplied by the preset time decay factor raised to its corresponding time point. The area under the curve at each fluctuation period is calculated using a numerical integration operator on the continuous-time curve of the product. All local maxima and minus points are extracted to form a multidimensional curve extremum set. For example, after decay calculation and integration peak finding, the extracted multidimensional curve extremum set contains four extrema: 1.2, 2.5, 0.8, and 3.1. For the obtained multidimensional curve extremum set, all extrema within the set are arithmetically summed, and the total sum is divided by the number of extrema to extract a global smoothing index. Substituting the values ​​1.2, 2.5, 0.8, and 3.1 into the previous example, we sum the four values ​​to obtain 7.6. Then, we divide 7.6 by the total number of extreme values, 4, to get the arithmetic mean of 1.9. This arithmetic mean of 1.9 is used as the global smoothing index, and the value of this global smoothing index is directly assigned to generate the material oxidation rate characteristic value of the current material. The advantage of this calculation logic is that by introducing a preset time decay factor for time-series weighting and extracting the extreme value set to obtain the mean, it can smooth out the drastic fluctuations in data caused by the instantaneous heating of the furnace.

[0030] Please see Figure 4 The specific steps of S3 are as follows: S301: Call the material oxidation rate feature value, read the discrete stage exposure time sequence in the process path library, perform element-wise multiplication based on the material oxidation rate feature value and the discrete stage exposure time sequence to obtain the node loss feature value set, and perform topological mapping according to the node number to establish a single-segment oxidation loss matrix; The generated material oxidation rate characteristic value is obtained from the computational center, and simultaneously, the underlying process database is accessed to read the pre-planned multi-stage discrete exposure duration sequence for the precious metal part. This discrete stage exposure duration sequence includes the continuous exposure seconds of the batch of material in the high-temperature and oxygen-rich environment during the first, second, and third processing stages. Element-wise multiplication is performed on the material oxidation rate characteristic value and this discrete stage exposure duration sequence, that is, the material oxidation rate characteristic value is multiplied by the exposure seconds of each discrete stage to obtain the theoretical oxide generation equivalent for each processing stage, thus forming a set of nodal loss characteristic values. For example, if the previously obtained material oxidation rate characteristic value is 1.9, and the read discrete stage exposure duration sequence includes 50 seconds for stage 1, 80 seconds for stage 2, and 30 seconds for stage 3, the nodal loss characteristic value set obtained after performing multiplication operations includes three values: 95, 152, and 57. Subsequently, based on the sequence of processing stages, i.e., the node numbers, a topological mapping is performed on each value in the node loss feature value set. These values ​​are then arranged in a row-major order into a 1x3 array to establish a single-segment oxidation loss matrix. After substituting the values, the resulting single-segment oxidation loss matrix contains row vectors with values ​​of 95, 152, and 57. The advantage of this operational logic is that by performing a direct multiplication mapping between the global feature values ​​and the independent duration of each discrete node, the absolute loss equivalent of each specific stage within the entire lifecycle can be quickly deconstructed. Accurate loss deconstruction is a prerequisite for building a high-quality processing service billing model and performance auditing.

[0031] S302: Call the single-segment oxidation loss matrix, read the local preset initial oxidation increment offset, perform time-series integral accumulation operation on the row vector to obtain the path process loss distribution vector, perform arithmetic summation on the path process loss distribution vector and the initial oxidation increment offset to establish the candidate path loss total; The system reads the newly established single-segment oxidation loss matrix from the internal storage medium and simultaneously reads the locally preset initial oxidation increment offset from the configuration file. This locally preset initial oxidation increment offset is a compensation value set to compensate for the weight of the extremely thin oxide layer naturally formed when the metal billet is exposed to ambient air before entering the heating furnace; it is set to 0.5 grams. The system extracts the row vectors contained within the single-segment oxidation loss matrix. Starting from the first element of each row vector, it performs a time-series integration operation to accumulate the cumulative loss at the completion of each stage, thus forming a path process loss distribution vector. For example, if the aforementioned row vector contains the values ​​95, 152, and 57, the first element accumulates to 95, the second element accumulates to 95 plus 152 equaling 247, and the third element accumulates to 247 plus 57 equaling 304, thereby generating a path process loss distribution vector containing the cumulative values ​​of these three stages: 95, 247, and 304. Next, the last maximum cumulative value in the path process loss distribution vector is extracted to represent the theoretical total loss of the processing path. This last maximum cumulative value is then arithmetically summed with the read local preset initial oxidation increment offset. The sum is used to establish the total candidate path loss under this specific processing plan. Substituting this into the numerical calculation, the last value of the path process loss distribution vector, 304, is added to the initial oxidation increment offset of 0.5 grams, yielding a total candidate path loss of 304.5 grams.

[0032] S303: Call the total loss of candidate paths to perform bubble sort, obtain the ascending list of loss values, retrieve the first and smallest value item from it, associate the identification code with the smallest value item, and generate a low-loss processing path. After multi-threaded concurrent execution of the aforementioned calculations to obtain the total cost of candidate paths under multiple scheduling plans, the scattered set of total cost values ​​for candidate paths is loaded into a memory sorting buffer. Bubble sort logic is then used to perform a loop comparison and position swapping on all candidate path cost values, moving smaller elements to the front of the list through layer-by-layer swaps, resulting in an ascending linked list completely sorted by cost value from smallest to largest. For example, if the total cost of candidate paths for three scheduling schemes is calculated to be 304.5 grams, 280.2 grams, and 315.8 grams, the ascending linked list generated after bubble sort will contain these values ​​in the order of 280.2 grams, 304.5 grams, and 315.8 grams. The smallest first value at the first position of this ascending linked list is retrieved directly using an addressing instruction. In this example, the retrieved smallest first value is 280.2 grams. Subsequently, the path identification code string corresponding to the minimum value item 280.2 grams is retrieved from the path planning index table. Data binding and association identification codes are then performed on the minimum value item and its corresponding path identification code string. The detailed scheduling parameter set pointed to by the identification code is formally packaged to generate a low-loss processing path to guide the actual actions of subsequent physical equipment. This enables the delivery of high-yield, low-loss, high-quality manufacturing and processing services to end customers, thereby enhancing the overall commercial competitiveness of the entire cloud manufacturing and processing service system.

[0033] Please see Figure 5 The specific steps of S4 are as follows: S401: For low-loss processing paths, collect the continuous torque state sequence and the set of surface reduction rate feature parameters, and perform time-series alignment to extract feature samples of the continuous torque state sequence and the set of surface reduction rate feature parameters according to the timestamps of the nodes in the low-loss processing path, and establish a dynamic mapping matrix. The control parameters of the determined low-loss processing path are retrieved. Simultaneously, a torque continuous state sequence is constructed by collecting spindle rotation resistance data every 10 milliseconds using a torque sensor mounted on the mill spindle. Furthermore, strip thickness variation data is simultaneously collected at the rolling exit using a thickness gauge to calculate a set of reduction rate characteristic parameters. Using the pre-set start and end timestamps of each processing pass in the low-loss processing path as a unified reference clock, a timing alignment operation based on the reference clock is performed on the torque continuous state sequence and the reduction rate characteristic parameter set. Torque data and reduction rate data with an absolute time difference of less than 5 milliseconds from the same timestamp are forcibly bound to the same data frame to extract feature samples, discarding unmatched data points. For example, at the timestamp of the first pass, the aligned spindle torque value is 1500 Nm, and the corresponding reduction rate value is 20%. These two values ​​are bound as one feature sample. The feature samples extracted from all passes of the low-loss processing path are aligned and stored in a two-dimensional array structure, with torque value as one column element and area reduction rate value as the other column element of the corresponding row, to establish a dynamic mapping matrix. The advantage of this operation logic is that the forced alignment and discarding of mismatched data based on timestamps eliminates data misalignment caused by sensor communication delays.

[0034] S402: Call the dynamic mapping matrix, extract the torque vector and deformation vector inside the dynamic mapping matrix, perform parametric fusion analysis on the torque vector and deformation vector to obtain the instantaneous work equivalent sequence, and perform smoothing and noise reduction on the instantaneous work equivalent sequence along the time axis to obtain the steady-state heat load vector. The system reads the constructed dynamic mapping matrix storage area and extracts the torque vector representing rotational resistance and the deformation vector (i.e., the set of reduction rates) representing the degree of cross-sectional deformation. For each set of torque and deformation vector elements aligned to the same timestamp, parametric fusion parsing is performed. The torque value and the reduction rate value are arithmetically multiplied to obtain a product. This product is then multiplied by a preset plastic work conversion coefficient of 0.85 to obtain the instantaneous work equivalent representing the local internal energy increment. The instantaneous work equivalents calculated from all feature samples are concatenated in chronological order to generate an instantaneous work equivalent sequence. For example, the extracted torque value of 1500 Nm and the reduction rate of 20% (0.2) are multiplied to obtain 300, which is then multiplied by the conversion coefficient of 0.85 to obtain the instantaneous work equivalent of 255 Joules for the current node. For this instantaneous work equivalent sequence, a continuous fluctuation segment with a duration of 20 seconds is extracted along the time axis. Within this segment, a moving average filtering algorithm is applied to perform smoothing and noise reduction. The filtering window length is set to 5 data points. The arithmetic mean of every 5 consecutive data points is calculated to replace the original value at the center point, filtering out high-frequency abrupt spikes caused by localized material hardening. The smoothed numerical array obtained after noise reduction is then saved as a steady-state heat load vector. The advantage of this operational logic is that it directly maps the change in internal energy by multiplying torque and deformation by a conversion coefficient, and combined with moving average noise reduction, it effectively eliminates the interference of mechanical vibration on thermal energy assessment.

[0035] Table 2: Analysis of kinetic parameters and work heat load; As shown in Table 2, by aligning the timestamps to obtain the torque and reduction rate, performing fusion analysis, and combining the conversion coefficients, the instantaneous work equivalents of different nodes can be obtained, thus providing the original input for generating a stable heat load sequence.

[0036] S403: Based on the steady-state heat load vector, collect the temperature drop rate parameter and read the specific heat capacity constant. Based on the thermal correlation characteristics between the temperature drop rate parameter and the specific heat capacity constant, construct the heat dissipation benchmark threshold. Compare the steady-state heat load vector with the heat dissipation benchmark threshold to extract the deviation set and generate the equipment material heat surplus state quantity. The system receives the output steady-state heat load vector and simultaneously uses an infrared temperature probe located in the cooling water spray zone to collect the temperature difference of the metal surface before and after passing through the cooling zone. Dividing this difference by the elapsed time yields the temperature drop rate parameter. The specific heat capacity constant of platinum at the current purity is retrieved from the local material property database as 130 joules per kilogram (Kelvin). Based on the thermodynamic correlation between the temperature drop rate parameter and the specific heat capacity constant, an arithmetic multiplication operation is performed between the temperature drop rate parameter and the specific heat capacity constant to calculate a baseline value of the heat that the material can theoretically dissipate per unit time, which serves as the heat dissipation baseline threshold. For example, if the temperature drop rate parameter obtained from the infrared temperature probe is 8 degrees Celsius per second, multiplying 8 degrees Celsius by the specific heat capacity constant of platinum (130) yields a heat dissipation baseline threshold of 1040 joules per second. After obtaining the heat dissipation baseline threshold, the heat load value at each time point in the previously obtained steady-state heat load vector is compared and subtracted from the heat dissipation baseline threshold of 1040 joules per second. If the heat load value is greater than the heat dissipation baseline threshold, the positive deviation value is extracted and recorded in the deviation set. For example, if the steady-state heat load at a certain time point is 1200 joules per second, subtracting 1040 joules per second yields a deviation of 160 joules per second, which is then included in the deviation set. The deviation set, composed of all positive differences obtained after traversing the entire sequence, is directly packaged and packaged to generate a thermal surplus state quantity of equipment materials used to characterize the degree of heat accumulation.

[0037] Please see Figure 6 The specific steps of S5 are as follows: S501: Call the low-loss processing path to extract the process time node distribution vector, call the equipment material heat surplus state quantity and preset recrystallization temperature threshold to perform over-limit state analysis to generate thermal overflow feature items, associate the time axis coordinates to assign high-level logic placeholders, and establish an over-limit state marker vector. The process invokes the low-loss processing path in the main memory and parses its instruction set, extracting the absolute time coordinates of the start and end of each processing step to form a process time node distribution vector. Next, it invokes the equipment material thermal surplus state quantity, which contains a large amount of positive deviation data, and extracts the extreme value data from this state quantity and performs over-limit state analysis against a preset recrystallization temperature threshold. This preset recrystallization temperature threshold is not a fixed constant, but rather identified by reading the laser etching attribute identifier on the surface of the material to be processed to determine the material grade. Then, it enters a preset material thermophysical constant mapping table for memory addressing, using the material grade for comparison and matching to extract the critical temperature value specific to the target material. For example, if the identifier identifies the current material as a specific alloy, the address comparison shows that the critical recrystallization temperature value of this alloy in the mapping table is 850 degrees Celsius. The peak temperature data in the equipment material thermal surplus state quantity is compared with this 850 degrees Celsius. When the peak temperature data is greater than 850 degrees Celsius, an over-limit condition is determined, and a thermal overflow characteristic item is generated. For each instant when a thermal overflow feature occurs, its corresponding time axis coordinate in the process time node distribution vector is retrieved. This time coordinate is then forcibly assigned a high-level logic placeholder (value 1) in a one-dimensional logic array of equal length to the time axis. Other time points that do not exceed the limit are assigned a low-level value (value 0), thus establishing a complete over-limit status marker vector describing the thermal over-limit time distribution. The advantage of this operational logic is that it uses a dynamic threshold with address matching combined with Boolean logic placeholder mapping, completely overcoming the limitations of rigid monitoring with a single threshold and achieving adaptive safety boundary calibration.

[0038] S502: Call the over-limit status flag vector, read the preset response delay period constant, and perform timing forward parsing to extract the early trigger timestamp for the coordinate position of the high-level logic placeholder. Configure the signal amplitude and signal pulse width for the early trigger timestamp and perform waveform fitting to generate the shutdown cooling trigger pulse. The system reads the out-of-limit status marker vector generated in memory in the previous stage, and simultaneously reads the preset response delay period constant from the safety configuration file. This constant represents the inherent physical hysteresis time required for the robotic arm and coolant pump to reach full power output from receiving a command; here, it is set to 3 seconds. The system iterates through the out-of-limit status marker vector, searching for all high-level logic placeholders assigned a value of 1, and extracts their corresponding timeline coordinates. For each high-level coordinate, a timing advance parsing is performed, subtracting the preset response delay period constant of 3 seconds from the time value represented by that coordinate, thus extracting an advance time coordinate, i.e., an advance trigger timestamp. For example, if a high-level logic placeholder is detected at the 120th second timeline coordinate, subtracting 3 seconds and advancing the timeline yields an advance trigger timestamp of 117 seconds. For each acquired advance trigger timestamp, the signal generator function is directly called to forcibly configure the signal amplitude of its output waveform to the standard control level of 24 volts and set the signal pulse width to be maintained for the action to 500 milliseconds. Square wave waveform fitting is performed on the set time domain and frequency domain parameters to generate a digital pulse control sequence containing the start level, maintenance width and turn-off level, that is, to generate the shutdown cooling trigger pulse used to drive the emergency stop and emergency cooling hardware.

[0039] S503: Call the shutdown cooling trigger pulse, obtain the basic execution timing chain, embed the shutdown cooling trigger pulse into the idle time slot of the basic execution timing chain according to the advance trigger timestamp clock cycle, obtain the hybrid scheduling instruction signal set, perform anti-collision verification to eliminate overlapping time windows, and obtain the production scheduling control sequence of precious metal processing services. The basic execution sequence list for the day is downloaded from the dispatch center. This list contains the predetermined start and end time distribution sequences for various routine operations such as feeding, rolling, and coiling. The previously generated shutdown cooling trigger pulse, containing high-frequency 24-volt square wave data, is invoked. Based on the clock cycle corresponding to the calculated advance trigger timestamp, the execution duration requirement of this shutdown cooling trigger pulse is forcibly inserted into an idle time slot in the basic execution sequence list where no routine tasks are scheduled. For example, if the calculated advance trigger timestamp is 117 seconds and the pulse width is 500 milliseconds, and a search of the basic execution sequence list reveals no other material movement actions with idle time slots between 116 and 118 seconds, then the shutdown cooling trigger pulse is directly embedded into this time period. After completing the initial embedding operation, a hybrid scheduling instruction signal set is acquired. Then, an anti-collision verification engine is invoked to perform a time-domain scan verification of the hybrid scheduling instruction signal set, checking whether any physical execution component is assigned two mutually exclusive action instructions within the same time window. If instruction overlap is found within the same time window, the lower-priority regular processing action instruction is forcibly removed, while the shutdown and cooling instruction is retained, thus eliminating the overlapping time window. After mine clearance and time slot optimization operations, the final safe and conflict-free sequence data is extracted to obtain the production scheduling control sequence for precious metal processing services output to the PLC controller. This marks the formal establishment of a fully automated and flexible processing service ecosystem encompassing multimodal perception monitoring, health status assessment, and intelligent anti-collision scheduling. This highly integrated processing service architecture achieves deep integration of software-defined hardware.

[0040] Table 3: Scheduling sequence anti-collision verification process table; As shown in Table 3, by searching for scheduling overlaps in each time segment and forcibly removing low-priority conflicting tasks, the final high-safety-level production scheduling control sequence can be effectively obtained.

[0041] Please see Figure 7 A smart production scheduling system for precious metal processing, comprising: The geometric feature mapping module collects the length, width, and diameter of the precious metal billet, performs a specific surface area mapping transformation on the geometric shape features of the precious metal billet, calculates the ratio between the transformation result and the mass of the precious metal billet, and generates the mass specific surface area. The oxidation rate fitting module calls the mass specific surface area, collects the furnace lining temperature and oxygen partial pressure inside the melting furnace, performs fitting calculations through a high-temperature oxidation kinetic model, calculates the oxidation reaction rate on the surface of the precious metal billet, and generates material oxidation rate characteristic values. The loss timing optimization module calls the material oxidation rate characteristic value, retrieves the exposure time from the preset process path library, performs timing loss accumulation calculation, and selects the path with the highest total loss from the preset process path library to determine the low loss processing path. The thermal balance state verification module collects motor torque and material reduction rate for low-loss processing paths and performs vector multiplication calculation. It then verifies the thermal balance by combining the temperature drop rate parameter and obtains the material thermal surplus state of the equipment. The scheduling pulse generation module reads the process time nodes of the target low-loss processing path, performs a comparison and judgment between the equipment material thermal surplus state and the recrystallization temperature threshold, embeds a shutdown cooling trigger pulse at the process time node, and generates a production scheduling control sequence.

[0042] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the described technical solutions.

Claims

1. A method for intelligent production scheduling in precious metal processing, characterized in that, Includes the following steps: S1: Collect the length, width and diameter of the precious metal billet, perform specific surface area mapping transformation based on the geometric shape characteristics of the precious metal billet, calculate the ratio between the transformation result and the mass of the precious metal billet, and generate the mass specific surface area. S2: Call the mass specific surface area, collect the furnace lining temperature and oxygen partial pressure inside the melting furnace, perform fitting calculation through the high-temperature oxidation kinetic model, calculate the oxidation reaction rate on the surface of the precious metal billet, and generate material oxidation rate characteristic value; S3: Call the material oxidation rate characteristic value, retrieve the exposure time from the preset process path library, perform time-series loss accumulation calculation, and select the path with the highest total loss from the preset process path library to determine the low-loss processing path. S4: For the low-loss processing path, collect the motor torque and material reduction rate and perform vector multiplication. Combine the temperature drop rate parameter to check the thermal balance and obtain the material thermal surplus state of the equipment. S5: Read the process time node of the low-loss processing path, perform a comparison and determination of the material thermal surplus state of the equipment and the recrystallization temperature threshold, embed a shutdown cooling trigger pulse at the process time node, and generate a production scheduling control sequence.

2. The intelligent production scheduling method for precious metal processing according to claim 1, characterized in that, The specific surface area includes surface roughness, material porosity, and geometric shape factor; the material oxidation rate characteristic values ​​include oxidation weight gain, oxide film thickness, and parabolic rate constant; the low-loss processing path includes process flow sequence, processing lead time, and station dwell time; the equipment material thermal surplus state quantity includes material deformation heat, interfacial frictional heat, and system heat loss; and the production scheduling control sequence includes production cycle time, start-up time, and shutdown cooling time.

3. The intelligent production scheduling method for precious metal processing according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Collect a set of three-dimensional coordinate point clouds of metal billet, filter outliers from the three-dimensional coordinate point cloud set, extract the outer edge contour points, calculate the extreme difference of the outer edge contour points in the coordinate axis projection direction, obtain the entity length, width and diameter values, and combine them to establish a spatial three-dimensional feature matrix. S102: Call the spatial three-dimensional feature matrix, calculate the entity length value, entity width value and entity diameter value based on the integral operator to generate the outer surface area value, calculate the internal volume space value based on the volume operator, divide the outer surface area value by the internal volume space value to obtain the surface area ratio mapping value. S103: The monitoring base sensor acquires the weight data of the metal billet, performs a division operation on the surface area ratio mapping value and the weight data of the metal billet to obtain a ratio operation number sequence, performs bit truncation and recombination on the ratio operation number sequence according to the preset precision verification floating-point base comparison value, and generates mass specific surface area.

4. The intelligent production scheduling method for precious metal processing according to claim 3, characterized in that, The specific steps of S2 are as follows: S201: Call the mass specific surface area, collect the furnace lining temperature distribution sequence and oxygen partial pressure value, perform normalization mapping on the mass specific surface area, furnace lining temperature distribution sequence and oxygen partial pressure value to obtain environmental feature vector, and establish furnace oxidation potential energy matrix in combination with preset activation energy constant. S202: Extract the row vectors of the furnace oxidation potential energy matrix to construct a multidimensional potential energy space, perform a product operation on the mass specific surface area and the furnace oxidation potential energy matrix to obtain the reaction intensity vector, extract the sliding window discrete change amount from the reaction intensity vector, and generate a basic reaction evolution rate sequence. S203: Call the basic reaction evolution rate sequence, perform continuous integration operation on the basic reaction evolution rate sequence and the preset time decay factor to extract the multidimensional curve extreme value set, perform arithmetic mean calculation on the multidimensional curve extreme value set to extract the global smoothing index, and generate material oxidation rate characteristic value.

5. The intelligent production scheduling method for precious metal processing according to claim 4, characterized in that, The extraction of discrete change in the sliding window for the reaction intensity vector refers to setting the sampling length and displacement step of the sliding window, truncating the reaction intensity vector according to the sampling length to obtain a vector subsequence, extracting the difference between the last value and the first value of the vector subsequence, dividing the difference by the sampling length to obtain the local rate of change, translating the sliding window according to the displacement step, and collecting multiple sets of local rate of change combinations to generate discrete change in the sliding window.

6. The intelligent production scheduling method for precious metal processing according to claim 4, characterized in that, The specific steps for S3 are as follows: S301: Call the material oxidation rate feature value, read the discrete stage exposure time sequence in the process path library, perform element-wise multiplication based on the material oxidation rate feature value and the discrete stage exposure time sequence to obtain the node loss feature value set, and perform topological mapping according to the node number to establish a single-segment oxidation loss matrix; S302: Call the single-segment oxidation loss matrix, read the local preset initial oxidation increment offset, perform time-series integral accumulation operation on the row vector to obtain the path process loss distribution vector, perform arithmetic summation on the path process loss distribution vector and the initial oxidation increment offset to establish the candidate path loss total. S303: Call the total loss of the candidate path to perform bubble sort, obtain the linked list of loss values ​​in ascending order, retrieve the first and smallest value item from it, associate the identification code with the smallest value item, and generate a low-loss processing path.

7. The intelligent production scheduling method for precious metal processing according to claim 6, characterized in that, The specific steps of S4 are as follows: S401: For the low-loss processing path, collect the torque continuous state sequence and the set of surface reduction rate feature parameters, and perform time-series alignment to extract feature samples of the torque continuous state sequence and the set of surface reduction rate feature parameters according to the timestamp of the low-loss processing path node to establish a dynamic mapping matrix. S402: Call the dynamic mapping matrix, extract the torque vector and deformation vector inside the dynamic mapping matrix, perform parametric fusion analysis on the torque vector and deformation vector to obtain the instantaneous work equivalent sequence, and perform smoothing and noise reduction on the instantaneous work equivalent sequence by truncating a segment along the time axis to obtain the steady-state heat load vector. S403: Based on the steady-state heat load vector, collect the temperature drop rate parameter and read the specific heat capacity constant. Construct a heat dissipation benchmark threshold based on the thermal correlation characteristics between the temperature drop rate parameter and the specific heat capacity constant. Compare the steady-state heat load vector with the heat dissipation benchmark threshold to extract the deviation set and generate the equipment material heat surplus state quantity.

8. The intelligent production scheduling method for precious metal processing according to claim 7, characterized in that, The specific steps of S5 are as follows: S501: Call the low-loss processing path to extract the process time node distribution vector, call the equipment material heat surplus state quantity and preset recrystallization temperature threshold to perform over-limit state analysis to generate heat overflow feature item, associate time axis coordinates to assign high-level logic placeholders, and establish over-limit state marker vector. S502: Call the over-limit state marker vector, read the preset response delay period constant and perform timing forward parsing on the coordinate position of the high-level logic placeholder to extract the early trigger timestamp, configure the signal amplitude and signal pulse width for the early trigger timestamp and perform waveform fitting to generate a shutdown cooling trigger pulse; S503: Invoke the shutdown cooling trigger pulse, obtain the basic execution timing list, embed the shutdown cooling trigger pulse into the idle time slot of the basic execution timing list according to the advance trigger timestamp clock cycle, obtain the hybrid scheduling instruction signal set, perform anti-collision verification to eliminate overlapping time windows, and obtain the production scheduling control sequence of precious metal processing services.

9. The intelligent production scheduling method for precious metal processing according to claim 8, characterized in that, The preset recrystallization temperature threshold is determined by parsing the attribute identifier of the material to be processed and performing addressing comparison in a preset material thermophysical constant mapping table to extract the critical temperature value of the target material.

10. A smart production scheduling system for precious metal processing, characterized in that, The system is used to implement the intelligent production scheduling method for precious metal processing according to any one of claims 1-9, the system comprising: The geometric feature mapping module collects the length, width, and diameter of the precious metal billet, performs a specific surface area mapping transformation on the geometric shape features of the precious metal billet, calculates the ratio between the transformation result and the mass of the precious metal billet, and generates the mass specific surface area. The oxidation rate fitting module calls the mass specific surface area, collects the furnace lining temperature and oxygen partial pressure inside the melting furnace, performs fitting calculations through a high-temperature oxidation kinetic model, calculates the oxidation reaction rate on the surface of the precious metal billet, and generates material oxidation rate characteristic values. The loss timing optimization module calls the material oxidation rate characteristic value, retrieves the exposure time from the preset process path library, performs timing loss accumulation calculation, and selects the path with the highest total loss from the preset process path library to determine the low loss processing path. The thermal balance state verification module collects motor torque and material reduction rate for the low-loss processing path and performs vector multiplication calculation. It then verifies the thermal balance by combining the temperature drop rate parameter and obtains the material thermal surplus state of the equipment. The scheduling pulse generation module reads the process time nodes of the target low-loss processing path, performs a comparison and determination of the material thermal surplus state of the equipment and the recrystallization temperature threshold, embeds a shutdown cooling trigger pulse at the process time node, and generates a production scheduling control sequence.

Images