Workpiece quality online analysis and production decision optimization model

Abstract

The invention discloses a workpiece quality online analysis and production decision optimization model, and belongs to the field of industrial production management, and the model comprises a spectrum generation module which is used for generating a one-dimensional process energy density spectrum on line, and the process energy density spectrum is formed by dynamically distributing weights for a plurality of process control variables and carrying out nonlinear superposition; the factor construction module is used for calculating a stable factor of each production takt point based on the process energy density spectrum so as to construct a takt characteristic field; the stability factor is determined according to the deviation degree of the current spectrum value relative to the target reference, the spectrum value change gradient of the adjacent beat points and the state logic values of the downstream buffer area and the downstream process; and the instruction splitting module is used for splitting the optimization decision instruction into a plurality of decision quantum packets. Accurate adaptation of the decision instruction and the production takt and dynamic resource allocation can be achieved, resource waste caused by a traditional fixed process is avoided, the utilization efficiency of production resources such as equipment and materials is improved, the production cost is reduced, and the flexibility of production management and control is enhanced.

Description

Technical Field

[0001] This invention relates to the field of industrial production management, and more specifically, to an online analysis model for workpiece quality and a production decision optimization model. Background Technology

[0002] Currently, quality analysis in industrial production largely relies on monitoring single variables or manual experience-based judgment, lacking the ability to systematically analyze the synergistic effects of multiple variables. Traditional methods often view the impact of each process parameter in isolation, failing to effectively uncover the deep correlations between parameters and their implicit mechanisms of action on quality indicators. This makes it difficult to quickly pinpoint the root cause when quality anomalies occur. At the same time, the production decision-making process is limited by the efficiency of manual data processing and the limitations of experience, resulting in decisions lacking data support and scientific validity. These decisions often lag behind the dynamic changes on the production floor and cannot respond promptly to quality risks caused by process fluctuations.

[0003] In production management processes such as resource scheduling and workflow planning, traditional models often employ fixed processes or static allocation strategies, failing to dynamically adjust resource allocation and process parameters based on real-time quality data. When parameter drift or equipment status fluctuations occur during production, existing decision support systems struggle to quickly generate targeted optimization solutions, resulting in insufficient stability in the production process and significant fluctuations in product yield. Furthermore, traditional troubleshooting often requires manual step-by-step investigation, which can take several hours, severely impacting production efficiency and increasing production costs and quality losses. Summary of the Invention

[0004] To address the problems existing in the prior art, the purpose of this invention is to provide an online workpiece quality analysis and production decision optimization model, which can achieve precise matching of decision instructions and production cycle and dynamic allocation of resources, avoid resource waste caused by traditional fixed processes, improve the utilization efficiency of production resources such as equipment and materials, reduce production costs, and enhance the flexibility of production management.

[0005] To solve the above problems, the present invention adopts the following technical solution:

[0006] Online workpiece quality analysis and production decision optimization model, including:

[0007] The spectrum generation module is used to generate a one-dimensional process energy density spectrum online. The process energy density spectrum is formed by dynamically assigning weights to multiple process control variables and performing nonlinear superposition.

[0008] The factor construction module calculates the stability factor for each production cycle point based on the process energy density spectrum to construct the cycle characteristic field. The stability factor is determined based on the deviation of the current spectrum value from the target benchmark, the gradient of spectrum value change at adjacent cycle points, and the state logic values ​​of the downstream buffer and downstream process.

[0009] The instruction splitting module is used to split the optimization decision instruction into multiple decision quantum packages and bind each decision quantum package to the future production tick point in the tick characteristic field where the stability factor exceeds a set threshold.

[0010] The reverse simulation module is used to perform reverse perturbation simulation on the decision quantum packet sequence of the bound beat point. Starting from the end of the sequence, it calculates the amount of modification to the process energy density spectrum after each decision quantum packet takes effect, and checks whether it will cause the stability factor of any beat point to fall below the set threshold.

[0011] The proxy game module is used to take the verified decision quantum package as a proxy and play a proxy game in the interval of the beat point where the stability factor exceeds the set threshold, so as to adjust its binding beat point and form an optimized sequence.

[0012] The beat synchronization module is used to synchronize the actual production beat with the optimization sequence. When the beat reaches the binding beat point of any decision quantum package, it executes the adjustment instructions contained therein.

[0013] Furthermore, the spectrum generation module includes:

[0014] For each process control variable, based on its current value relative to its own process constraint window and its acceleration of change, the instantaneous influence potential sequence changing with time is calculated;

[0015] Based on the instantaneous influence potential sequence of each process control variable, a temporal entanglement matrix is ​​constructed, where each element of the matrix is ​​determined based on the waveform similarity and phase difference of each pair of instantaneous influence potential sequences;

[0016] The instantaneous influence potential sequence of each process control variable is convolved with the dynamic convolution kernel constructed by the row vectors corresponding to the process control variables in the temporal entanglement matrix to generate the original density flow corresponding to each variable.

[0017] At each moment, the dominant flow with the largest gradient among all the primary density flows is identified, and the energy of each primary density flow at the current moment is focused onto the characteristic frequency envelope of the dominant flow, which is then collapsed to generate a one-dimensional process energy density spectrum.

[0018] Furthermore, the factor construction module includes:

[0019] Based on the difference between the current value and the target reference value of the process energy density spectrum, and combined with the local frequency characteristics of the process energy density spectrum, the influence cone projection vector is calculated and generated. Each component of the influence cone projection vector represents the expected influence amplitude of the current spectral value deviation on future continuous beat points.

[0020] Based on the curvature and direction of change of the process energy density spectrum near the beat point, the inertial damping factor at the beat point is calculated.

[0021] Furthermore, the factor construction module is specifically used for:

[0022] Based on the physical occupancy status of the downstream buffer and the ready signal of the downstream process, the pressure transmission coefficient is calculated using the pressure transmission model.

[0023] Using the stability factor field distribution obtained from the previous iteration as the initial field distribution, the influence cone projection vector as the disturbance source, the inertial damping factor as the field medium absorption coefficient, and the pressure transmission coefficient as the boundary driving force, the stability factor at the current beat point is obtained and the stability factor field distribution is updated through field iterative equilibrium calculation.

[0024] Furthermore, the instruction splitting module includes:

[0025] The decision instructions are analyzed according to different adjustment dimensions. The potential impact amplitude is determined based on the adjustment magnitude and parameter sensitivity coefficient of each dimension. The action span is determined by combining the physical time window required for each dimension to take effect. In this way, multiple decision quantum packages with potential impact amplitude and action span attributes are generated.

[0026] For each candidate beat point in the beat characteristic field whose stability factor exceeds a set threshold, its dynamic affinity to each decision quantum packet is calculated. The dynamic affinity is determined by the matching degree between the stability factor margin of the beat point and the influence potential amplitude of the decision quantum packet, as well as the degree of agreement between the action span of the decision quantum packet and the future process stability window of the beat point predicted based on the process energy density spectrum.

[0027] Furthermore, the instruction splitting module is specifically used to perform the following binding operations:

[0028] Using the impact potential amplitude of each decision quantum package as the bidding power and the dynamic affinity of each tick point to each decision quantum package as the acceptance willingness, a multi-round two-way bidding matching is performed to determine the final tick point for binding each decision quantum package.

[0029] Furthermore, the reverse simulation module includes:

[0030] For each bound beat point, a characteristic perturbation wave is generated for the decision quantum packet. The initial amplitude of the characteristic perturbation wave is determined by the influence potential amplitude of the decision quantum packet, the wavelength is determined by its action time span, and it propagates attenuated along the time axis with the bound beat point of the decision quantum packet as the emission origin.

[0031] On the reverse time axis starting from the end of the sequence, the characteristic perturbation waves of all decision quantum packets are superimposed, and their interference effect is determined according to the phase of each characteristic perturbation wave, generating an interference superposition waveform with a comprehensive perturbation amplitude.

[0032] Furthermore, the reverse simulation module is also used for:

[0033] The combined disturbance amplitude of the interferometric superimposed waveform is mapped to a modification risk field of the process energy density spectrum at the corresponding tick point through the risk transfer function. The modification risk field includes the expected amount and uncertainty range of the spectrum value modification.

[0034] The modified risk field is used as a potential disturbance source and substituted into the beat characteristic field model for prospective simulation. The stability factor of the relevant beat point is re-estimated, and it is verified whether the re-estimated stability factor is lower than the set threshold.

[0035] Furthermore, the proxy game-playing module includes:

[0036] The intra-sequence conflict coefficients between each decision quantum bag that has been verified are calculated. The conflict coefficients are determined based on the temporal proximity of their binding beat points and the effect interference of the reverse perturbation simulation prediction. The migration intention is calculated for each decision quantum bag. The migration intention is positively correlated with the sum of its conflict coefficients and negatively correlated with the stability factor margin of its current binding beat point.

[0037] Based on migration intention and the potential magnitude of the influence of the decision quantum package, calculate the dynamic bargaining power of each decision agent, and assign game strategies to each decision agent based on this power;

[0038] Within the interval of beat points where the stability factor exceeds the set threshold, multiple rounds of negotiation are executed according to the game strategy. For each round of negotiation, the proposed beat point change is verified by calling the reverse perturbation simulation. The verified changes are recorded as temporary binding protocols, and an optimization decision sequence is formed based on the final set of protocols.

[0039] Furthermore, the beat synchronization module includes:

[0040] For each decision quantum packet in the optimization decision sequence, a dynamic countdown bridge is constructed. The bridge takes its bound tick point as the theoretical zeroing time and dynamically adjusts its elapsed speed according to the deviation rate between the monitored actual production tick rate and the theoretical tick rate. At the same time, the execution critical window around the zeroing point is calibrated.

[0041] When the dynamic countdown bridge enters its execution critical window, it verifies the real-time stability factor of the target tick point and the current state deviation of the relevant process equipment, and conditionally compiles the adjustment instructions contained in the decision quantum package based on the verification results, and issues the compiled instructions when the bridge returns to zero.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] (1) This scheme uses multivariate correlation analysis and process energy density spectrum construction to accurately discover the implicit correlation between process parameters and quality indicators, shorten the time for locating abnormal causes from hours to minutes, greatly improve the efficiency of quality problem investigation, and reduce quality loss caused by one-sided analysis.

[0044] (2) This solution replaces the traditional manual decision-making and static planning mode by splitting decision instructions, binding cycle time, and optimizing through multiple rounds of negotiation. This enables the dynamic generation and precise execution of decision suggestions, improves the speed of decision response, ensures that production adjustments are timely adapted to process fluctuations, and avoids production risks caused by decision lag. (3) This solution avoids process fluctuations that may be caused by decision execution in advance by calculating stability factors, simulating reverse perturbations, and performing forward-looking simulation verification. This effectively suppresses the fluctuation range of yield and significantly improves the stability of the production process compared to the traditional production mode, ensuring the consistency and reliability of product quality.

[0045] (4) Based on the matching of the cycle characteristic field and dynamic affinity, this solution achieves accurate adaptation of decision instructions and production cycle and dynamic allocation of resources, avoids resource waste caused by traditional fixed processes, improves the utilization efficiency of production resources such as equipment and materials, reduces production costs, and enhances the flexibility of production management. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating the workflow between the various modules in the online workpiece quality analysis and production decision optimization model of this invention. Detailed Implementation

[0048] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0049] Please see Figure 1 The online analysis and production decision optimization model for workpiece quality includes:

[0050] The spectrum generation module is used to generate a one-dimensional process energy density spectrum online. The process energy density spectrum is formed by dynamically assigning weights to multiple process control variables and performing nonlinear superposition.

[0051] The factor construction module calculates the stability factor for each production cycle point based on the process energy density spectrum to construct the cycle characteristic field. The stability factor is determined based on the deviation of the current spectrum value from the target benchmark, the gradient of spectrum value change at adjacent cycle points, and the state logic values ​​of the downstream buffer and downstream process.

[0052] The instruction splitting module is used to split the optimization decision instruction into multiple decision quantum packages and bind each decision quantum package to the future production tick point in the tick characteristic field where the stability factor exceeds a set threshold.

[0053] The reverse simulation module is used to perform reverse perturbation simulation on the decision quantum packet sequence of the bound beat point. Starting from the end of the sequence, it calculates the amount of modification to the process energy density spectrum after each decision quantum packet takes effect, and checks whether it will cause the stability factor of any beat point to fall below the set threshold.

[0054] The proxy game module is used to take the verified decision quantum package as a proxy and play a proxy game in the interval of the beat point where the stability factor exceeds the set threshold, so as to adjust its binding beat point and form an optimized sequence.

[0055] The beat synchronization module is used to synchronize the actual production beat with the optimization sequence. When the beat reaches the binding beat point of any decision quantum package, it executes the adjustment instructions contained therein.

[0056] The spectrum generation module is also used to perform the following steps:

[0057] Step 11: For each process control variable, based on its current value's position relative to its own process constraint window and its acceleration of change, calculate the instantaneous influence potential sequence that changes over time. The specific operation is as follows:

[0058] To quantify the dynamic impact of each process control variable on the process energy density spectrum at different times, it is necessary to calculate the instantaneous impact potential sequence of each process control variable over time. This calculation process is based on the relative position of the current value of the process control variable with its own process constraint window, while incorporating the variable's acceleration to construct a quantitative index that reflects the intensity of the variable's dynamic effect. The process constraint window represents the normal operating range of the process control variable; the closer the current value is to the window boundary, the more significant its potential impact on process stability. The acceleration reflects the drastic degree of variable change; a larger acceleration indicates more drastic fluctuations in the variable's state and a stronger instantaneous impact on the process. Based on these two parameters, the instantaneous impact potential is calculated using a nonlinear function, the expression of which is: ;

[0059] First, the influence weight of the variable's static position is obtained by normalizing the deviation of the current value from the center of the constraint window. Then, the influence of dynamic changes is quantified by combining the absolute value of the acceleration. Both are weighted by coefficients. , The fusion forms a comprehensive instantaneous influence potential, where, Let represent the instantaneous influence potential of the i-th process control variable at time t. Let be the current value of the variable at time t. and These are the lower and upper limits of its process constraint window, respectively. To constrain the center value of the window, Let be the acceleration of the change of this variable at time t. , These are weighting coefficients preset based on process characteristics. Used to adjust the effects of static position. This calculation process is used to assess the dynamic effects of acceleration. It yields a sequence of instantaneous influence potentials for each process control variable as it changes continuously over time.

[0060] In addition, for different types of processes, , The possible values ​​differ; please refer to the following for details:

[0061] Step 12: Construct a temporal entanglement matrix based on the instantaneous influence potential sequence of each process control variable. Each element of the matrix is ​​determined based on the waveform similarity and phase difference of each pair of instantaneous influence potential sequences. The specific operation is as follows:

[0062] To characterize the correlation between the instantaneous influence potentials of different process control variables over time, a temporal entanglement matrix needs to be constructed based on the instantaneous influence potential sequences of each variable. This matrix quantifies the dynamic correlation characteristics between the instantaneous influence potential sequences of any two process control variables, providing data support for the subsequent construction of dynamic convolution kernels. Each element in the matrix corresponds to the temporal entanglement strength of a pair of process control variables, and its value is determined by the waveform similarity and phase difference of the instantaneous influence potential sequences of these two variables. Waveform similarity reflects the consistency of the changing trends of the two sequences; higher similarity indicates that the influence patterns of the two variables are more similar. Phase difference reflects the synchronicity of the two sequences on the time axis; smaller phase difference indicates that the peak or trough of the influence of the two variables occurs closer together. Based on these two correlation characteristics, the matrix elements are calculated using a weighted fusion formula: ;

[0063] in The weighting coefficient, with a value range of [0,1], is used to balance the influence weights of waveform similarity and phase difference. It is determined based on the correlation characteristics between process variables. The derivation logic of this formula is that waveform similarity... The value range is [0,1], which directly represents the degree of fit of the sequence trend, while the phase difference... The cosine value also ranges from [-1, 1]. After processing, it can characterize temporal synchronicity. By combining the two through linear weighting, the temporal correlation strength of the instantaneous influence potential sequence of the two variables can be comprehensively and quantitatively reflected, thus constructing a complete temporal entanglement matrix. In the formula, This represents the element values ​​corresponding to the i-th and j-th process control variables in the temporal entanglement matrix. Let the waveform similarity be the instantaneous influence potential sequence of the i-th and j-th process control variables. The phase difference between the two sequences is represented by the phase difference. Through this construction process, the temporal entanglement matrix can clearly show the temporal correlation distribution of all process control variables on the instantaneous influence potential.

[0064] Step 13: Convolve the instantaneous influence potential sequence of each process control variable with the dynamic convolution kernel constructed from the row vectors corresponding to the process control variables in the temporal entanglement matrix to generate the original density flow corresponding to each variable. The specific operation is as follows:

[0065] To integrate the temporal correlations between process control variables into the representation of the influence of a single variable, the instantaneous influence potential sequence of each process control variable needs to be convolved with a dynamic convolution kernel constructed based on a temporal entanglement matrix. This generates the native density flow corresponding to each variable. The construction of the dynamic convolution kernel is based on the row vector corresponding to the process control variable in the temporal entanglement matrix. Each element in the row vector represents the temporal entanglement strength of that variable with all other variables. After normalization, this row vector is used as the core weight parameter of the convolution kernel, enabling the kernel to dynamically reflect the degree of correlation influence of other variables on the current variable. The purpose of the convolution operation is to incorporate the temporal correlation information of other variables through a weighted summation using a sliding window. Information is integrated into the instantaneous influence potential sequence of the current variable, thereby correcting the influence representation of a single variable and making it more consistent with the actual process scenario of multi-variable synergy. During the convolution process, the window size of the dynamic convolution kernel is determined according to the time interval of the process cycle, and the weight coefficient is determined by the row vector elements of the temporal entanglement matrix. That is, the higher the temporal entanglement strength with the current variable, the larger the corresponding weight coefficient, and the more significant the correction effect on the instantaneous influence potential sequence of the current variable. Through this convolution operation, the instantaneous influence potential sequence of each process control variable is transformed into a native density stream that integrates multi-variable correlation information. This native density stream not only retains its own temporal change characteristics, but also integrates the dynamic influence of other related variables.

[0066] Step 14: At each time step, identify the dominant flow with the largest gradient among all primary density flows, and focus the energy of each primary density flow at the current time onto the characteristic frequency envelope of the dominant flow, then collapse it to generate a one-dimensional process energy density spectrum. The specific operation is as follows:

[0067] To achieve effective fusion of multivariate primary density flows and generate a single-dimensional process energy density spectrum that comprehensively characterizes the overall energy state of the process, it is necessary to first identify the dominant flow with the largest gradient among all primary density flows at each time step. The gradient of a primary density flow reflects its rate of energy change, and the dominant flow with the largest gradient means that the combination of variables contributes most significantly to the process energy state at the current time step and is a factor affecting process stability. After identifying the dominant flow, its characteristic frequency envelope needs to be extracted. The characteristic frequency envelope is obtained by performing spectral analysis on the dominant flow and can reflect the main frequency range and amplitude variation law of the dominant flow's energy distribution, thus representing the energy characteristics of the dominant flow. Subsequently, the energy of all primary density flows at the current time step is focused on the dominant flow. On the characteristic frequency envelope of the guiding flow, this process is achieved through energy redistribution: based on the temporal entanglement intensity of each primary density flow and the dominant flow, the energy of the non-dominant flow is proportionally mapped to the frequency range corresponding to the characteristic frequency envelope of the dominant flow, while preserving the energy distribution characteristics of the dominant flow itself, thus achieving the synergistic focusing of multiple energy flows. This energy focusing process is essentially the elimination of redundancy and core reinforcement of multivariate information. Through the collapse mechanism, the multidimensional primary density flows are integrated into a single-dimensional process energy density spectrum. This single-dimensional spectrum can not only comprehensively reflect the dynamic correlation influence of all process control variables, but also simplify the processing dimensions of subsequent quality analysis and decision optimization, making the quantitative analysis and anomaly identification of process energy state more efficient and accurate.

[0068] In a preferred embodiment of the present invention, the factor construction module is further configured to perform the following steps:

[0069] Step 21: Based on the difference between the current value and the target reference value of the process energy density spectrum, and combined with the local frequency characteristics of the process energy density spectrum, calculate and generate the influence cone projection vector. Each component of the influence cone projection vector represents the expected influence amplitude of the current spectral value deviation on future continuous cycle points. The specific operation is as follows:

[0070] To predict the ongoing impact of the current process energy density spectrum deviation on subsequent production processes, an influence cone projection vector needs to be generated. The purpose is to quantify the propagation effect of the current spectral deviation over time, providing a forward-looking basis for calculating the stability factor. This vector is calculated based on the difference between the current process energy density spectrum value and the target baseline value. This difference directly reflects the degree of deviation between the current process state and the ideal state, serving as the initial variable affecting the stability of subsequent cycle times. Simultaneously, the local frequency characteristics of the process energy density spectrum are considered. These characteristics characterize the fluctuations of the current deviation, with different frequency components corresponding to varying attenuation rates during deviation propagation. High-frequency fluctuations typically attenuate faster, while low-frequency fluctuations have a more persistent impact. Based on these two key parameters, the influence cone projection vector is calculated using the following formula: ;

[0071] in Let m be the index vectors for the next m consecutive beats. The derivation logic of this formula originates from the physical laws of deviation propagation. First, let's consider the deviation of the current spectral value... As a benchmark for the intensity of influence, multiplied by the local frequency characteristic weight. Correct the propagation attenuation characteristics, and then use an exponential function. The simulation demonstrates the attenuation effect of the influence over time. The greater the time distance between the future tick point and the current moment (i.e., the larger the vector index), the smaller the influence amplitude. This aligns with the gradual dissipation of deviation effects in actual processes. In the formula... Let be the projection vector of the influence cone at time t. , The current value of the process energy density spectrum at time t. As the target baseline value, The local frequency feature weight at time t is obtained from local spectral analysis of the spectrum, and its value ranges from [0,1]. The attenuation coefficient is preset according to the process type, which controls the attenuation rate. m is the preset number of future consecutive beat points. Through this calculation process, each component of the cone projection vector can accurately characterize the expected influence amplitude of the current spectral deviation on each future consecutive beat point, providing key disturbance source data for subsequent field iterative equilibrium calculation.

[0072] Step 22: Based on the curvature and direction of change of the process energy density spectrum near the beat point, calculate the inertial damping factor at the beat point. The specific operation is as follows:

[0073] To quantify the inertial constraints during the change of the process energy density spectrum and avoid misjudgments in stability factor calculation due to instantaneous fluctuations, an inertial damping factor needs to be calculated based on the spectral variation characteristics near the beat point. This factor simulates the ability of the process system's own inertia to suppress disturbances, making the stability factor calculation more closely reflect the dynamic response characteristics of the actual process. The calculation process uses the curvature of the process energy density spectrum near the beat point as the core parameter. The curvature reflects the severity of the spectral value change; the larger the absolute value of the curvature, the steeper the spectral value change near the beat point, the more severe the fluctuation of the process state, and the greater the required inertial damping of the system. Simultaneously, the persistence of the spectral value change direction is considered. This parameter is obtained by judging whether the spectral value change direction is consistent across multiple consecutive beat points. The stronger the persistence of the direction, the more stable the trend of the process state change, the weaker the inertial constraints of the system, and the damping factor can be appropriately reduced. Based on these two parameters, the inertial damping factor is calculated using the following formula: ;

[0074] in Given a preset maximum curvature threshold, the derivation logic of this formula is as follows: first, by varying the absolute value of the curvature... The magnitude of the quantifiable fluctuation is multiplied by the directional persistence coefficient. After correcting the inertial characteristics, the value is normalized by dividing by the maximum curvature threshold to ensure that the damping factor always falls within the [0,1] range, thus preventing the damping effect from becoming unbalanced due to excessive curvature. In the formula... Let be the inertial damping factor at the beat point at time t. Let be the curvature of the process energy density spectrum change near the beat point at time t; is the persistence coefficient of the spectral value change direction near the beat point at time t, with a value range of [0,1]. It is 1 when the direction is completely continuous and 0 when there is no persistence. The maximum curvature threshold is determined based on historical process data. The inertial damping factor obtained through this calculation process can accurately characterize the process system's ability to suppress current fluctuations due to inertia.

[0075] Step 23: Based on the physical occupancy status of the downstream buffer and the ready signal of the downstream process, calculate the pressure transmission coefficient using the pressure transmission model. The specific operation is as follows:

[0076] To fully consider the impact of upstream and downstream process coordination constraints on the stability of the current production cycle, a pressure transmission coefficient needs to be calculated using a pressure transmission model. This coefficient transforms the state feedback from downstream processes into the boundary driving force for calculating the current cycle stability factor, ensuring that the stability factor reflects the coordinated stability of the entire production chain. The pressure transmission coefficient is calculated based on two key indicators: the physical occupancy status of the downstream buffer and the readiness signal of the downstream process. The physical occupancy status of the downstream buffer is quantified by the occupancy rate; a higher occupancy rate indicates a weaker capacity of the downstream process to handle the current process, creating reverse pressure and affecting the process adjustment space of the current cycle. The readiness signal of the downstream process is a logical value. When the signal is valid, it indicates that the process is ready, alleviating downstream pressure; when the signal is invalid, it indicates that the process is not ready, increasing reverse pressure. Based on these two indicators, the pressure transmission coefficient is calculated using the following formula: ;

[0077] The derivation logic of the formula originates from the pressure transmission mechanism of upstream and downstream processes, firstly by considering the buffer occupancy rate. As a positive pressure source, the higher the occupancy rate, the greater the pressure. The inverse logic value (1-r(t)) of the downstream process's ready signal is then used as an auxiliary pressure indicator; this value is 1 when not ready, further amplifying the pressure. This is further amplified by weighting coefficients. By balancing the influence weights of the two indicators, a coefficient that comprehensively reflects the intensity of downstream pressure transmission is finally obtained. In the formula, is the pressure conduction coefficient at time t, with a value range of [0,1]. This is a weighting coefficient, preset according to the characteristics of the process link, with a value range of [0,1]. The physical occupancy rate of the downstream buffer at time t, with a value range of [0,1], is obtained by the ratio of the actual occupied capacity of the buffer to the maximum capacity. The ready signal for the downstream process at time t is represented by the logical value 1, which indicates that the process is ready and 0 indicates that it is not ready. Through this calculation process, the pressure transmission coefficient can accurately quantify the constraint strength of the downstream process on the stability of the current cycle time.

[0078] Step 24: Using the stability factor field distribution obtained in the previous iteration as the initial field distribution, and taking the influence cone projection vector as the disturbance source, the inertial damping factor as the field medium absorption coefficient, and the pressure transmission coefficient as the boundary driving force, the stability factor at the current beat point is obtained and the stability factor field distribution is updated through field iterative equilibrium calculation. The specific operations are as follows:

[0079] To comprehensively consider the synergistic effects of the above factors, calculate the stability factor at the current production cycle point, and construct a continuous cycle characteristic field, a field iterative equilibrium calculation method is required. This method simulates the interaction of various influencing factors in a virtual field space, obtaining a quantitative index reflecting the overall stability of the process through iterative convergence. The calculation process uses the stability factor field distribution obtained in the previous iteration as the initial field distribution, ensuring the continuity and temporal correlation of the cycle characteristic field and avoiding numerical abrupt changes caused by a single calculation. Simultaneously, the influence cone projection vector generated in step 21 is used as the disturbance source in the field space. This vector carries the expected impact of the current spectral deviation on future cycles and is a factor that triggers field fluctuations. The inertial damping factor calculated in step 22 is used as the absorption coefficient of the field medium to simulate the process system's own ability to suppress disturbances, reducing the impact of unstable fluctuations on the stability factor. The pressure transmission coefficient obtained in step 23 is used as the boundary driving force of the field, reflecting the constraint effect of downstream links on the current field state. Based on the above settings, the formula for field iterative equilibrium is: ;

[0080] After multiple iterations until The derivation of this formula is based on the energy balance principle in field theory, and the initial field distribution... Base state, disturbance source Provide energy input for field fluctuations, subtracting the absorption coefficient. After the suppression, the boundary driving force is superimposed. The resulting change in the field gradient, through multiple iterations, gradually brings the field energy to equilibrium. The field value at this point is the stability factor at the current beat point in space x. In the formula, Let x be the stability factor at position x at time t. The initial stability factor field distribution at time t-1, The gradient of the stabilizing factor field at time t-1. The preset convergence threshold is set according to the process accuracy requirements, and is usually [value missing]. In terms of scale, after obtaining the stability factor of the current beat point, it needs to be updated into the overall stability factor field distribution to form a beat characteristic field covering all production beat points. This field distribution can not only intuitively reflect the stability differences of each beat point, but also provide characteristic basis for the subsequent splitting and binding of decision instructions, and at the same time provide a new initial field distribution for the next iteration calculation, ensuring the dynamic adaptability of the entire system.

[0081] In a preferred embodiment of the present invention, the instruction splitting module is further configured to perform the following steps:

[0082] Step 31: Analyze the decision instructions according to different adjustment dimensions, determine the potential impact magnitude based on the adjustment range and parameter sensitivity coefficient of each dimension, and determine the action span by combining the physical time window required for each dimension to take effect. In this way, generate multiple decision quantum packages with potential impact magnitude and action span attributes. The specific operation is as follows:

[0083] First, the decision instructions need to be analyzed according to different dimensions of process adjustment. These adjustment dimensions typically cover aspects directly related to production quality, such as correction of process parameter setpoints, adjustment of equipment operating status, and optimization of material supply parameters. Each dimension corresponds to an independent process influence path. For each analyzed adjustment dimension, the potential impact amplitude needs to be determined by combining its adjustment magnitude and parameter sensitivity coefficient. The adjustment magnitude directly reflects the strength of the change in process status caused by the instruction in that dimension. The parameter sensitivity coefficient is obtained based on historical data statistics and characterizes the degree of influence of parameter changes in that dimension on quality indicators. The product of the two is normalized to form the potential impact amplitude, quantifying the actual effectiveness of the instruction in that dimension. At the same time, the action span needs to be determined according to the physical time window required for each adjustment dimension to take effect. The physical time window is determined by actual production constraints such as the response characteristics of process equipment and the stabilization cycle of material status. For example, temperature adjustment requires a certain amount of time to reach a stable value, and the number of production cycles corresponding to this stabilization process is its action span. Based on the above analysis, the potential impact amplitude is calculated using the following formula: ;

[0084] Based on the quantitative requirements of the command's effect intensity, this formula first adjusts the amplitude. Normalization is performed by constraining the range of process parameters to eliminate dimensional differences between parameters of different dimensions, and then multiplied by a parameter sensitivity coefficient. Emphasizing the influence weight of highly sensitive dimensions ensures that the potential influence magnitude objectively reflects the actual effect of the instruction. In the formula, Let d be the magnitude of the potential impact of the dth adjustment dimension, with a value range of [0,1]. The parameter sensitivity coefficient for the d-th dimension is obtained from the correlation analysis of historical quality data and process parameters, and its value ranges from [0,1]. Let d be the adjustment range for the d-th dimension; and These are the upper and lower limits of the process constraints for the d-th dimension parameter, respectively. Through the above process, each adjustment dimension is transformed into a decision quantum package containing the magnitude of the impact potential and the span of the effect. These quantum packages retain the optimization objective of the original decision instruction and also possess the key features that match the production cycle point.

[0085] Step 32: For each candidate beat point in the beat characteristic field whose stability factor exceeds a set threshold, calculate its dynamic affinity to each decision quantum packet. The dynamic affinity is determined by the matching degree between the beat point's stability factor margin and the magnitude of the decision quantum packet's influence potential, as well as the degree of agreement between the decision quantum packet's effect span and the future process stability window of the beat point predicted based on the process energy density spectrum. The specific operation is as follows:

[0086] First, production cycle points whose stability factor exceeds a set threshold are selected as candidate cycle points from the cycle point characteristic field. A stability factor exceeding the threshold indicates that the process state at that cycle point possesses a certain degree of stability, capable of withstanding disturbances from decision commands without causing quality anomalies, and serving as a reliable carrier for command execution. For each candidate cycle point and each decision quantum packet, the calculation of dynamic affinity needs to consider two dimensions: one is the matching degree between the stability factor margin of the cycle point and the influence potential amplitude of the decision quantum packet. The stability factor margin is the difference between the actual stability factor of the candidate cycle point and the set threshold; a larger margin indicates that the cycle point can accommodate a greater number of disturbances from decision commands. The stronger the perturbation, the better the matching degree is obtained by nonlinear mapping the ratio of the two factors, ensuring that the potential impact amplitude does not exceed the carrying capacity of the tick point. Secondly, the matching degree is determined by the range of action of the decision quantum packet and the degree of agreement between the tick point and the future process stability window predicted based on the process energy density spectrum. The process stability window is obtained by predicting the trend of the process energy density spectrum, representing the time range within which the process state remains stable for multiple consecutive ticks after the tick point. A higher degree of agreement indicates a better match between the action period of the decision quantum packet and the process stability period, making it easier to achieve the expected effect after command execution. Based on these two dimensions, the dynamic affinity is calculated using the following formula: ;

[0087] Based on the dual constraints of adaptability, this formula first quantifies the adaptability of the cycle point to the instruction disturbance, avoiding process fluctuations caused by the instruction intensity exceeding the capacity. Then, it quantifies the adaptability of the time dimension, ensuring that the instruction action cycle matches the process stability cycle. Finally, it uses weighting coefficients... To balance the influence of both factors and ensure that affinity can fully reflect the rationality of the match, the formula is as follows: For the first The candidate beat point for the first The dynamic affinity of each decision quantum packet, with values ​​ranging from [0,1]; For the first Stability factor for each candidate beat point Set a threshold for the stability factor. For the first The potential magnitude of the impact of a decision quantum package For the first The scope of the effect of a decision quantum package For the first The future process stability window for each candidate beat point The weighting coefficient is preset according to the process characteristics, and its value range is [0,1]. Through this calculation process, the degree of fit between each candidate tick point and each decision quantum package is precisely quantized.

[0088] Step 33: Using the impact potential amplitude of each decision quantum package as the bidding power and the dynamic affinity of each tick point to each decision quantum package as the acceptance willingness, perform multiple rounds of two-way bidding matching to determine the final tick point bound to each decision quantum package. The specific operation is as follows:

[0089] To achieve optimal matching between decision quantum packages and candidate tick points, a multi-round bidirectional bidding matching mechanism needs to be constructed based on dynamic affinity and influence potential amplitude. The aim is to maximize the overall decision execution effect while satisfying the compatibility requirements of both, ensuring that instruction execution conforms to the carrying capacity of individual tick points while also optimizing overall production quality. In this mechanism, the influence potential amplitude of each decision quantum package is defined as its bidding ability. A higher influence potential amplitude indicates a greater potential contribution of the quantum package to process optimization, giving it a stronger competitive advantage in the bidding process and allowing it to prioritize tick points with higher compatibility. The dynamic affinity of each candidate tick point to each decision quantum package is defined as its acceptance willingness. A higher dynamic affinity indicates a better compatibility between the tick point and the corresponding decision quantum package, a lower risk of process fluctuations after executing the instruction, and a greater tendency to accept the decision quantum package. The bidding matching process is conducted in multiple iterations. In the first round of bidding, each decision quantum package sends a matching request to all candidate tick points based on its own bidding ability. The received requests are sorted according to acceptance willingness, and the decision quantum package with the highest acceptance willingness is selected to form a temporary matching relationship. Then, the next round of bidding begins. Decision quantum packages that have not been matched adjust their bidding strategies according to the acceptance willingness of the remaining candidate beat points. Unsaturated candidate beat points continue to receive matching requests and are filtered. Unsaturated candidate beat points can be represented by beat points that can carry multiple decision quantum packages. After each round of matching, the rationality of the temporary matching relationship needs to be verified. If the matching of a decision quantum package with a beat point will lead to insufficient subsequent carrying capacity of the beat point, or if there is a better matching combination, the current temporary relationship is terminated and the next round of iteration begins. The bidding process stops when all decision quantum packages have found a matching beat point, or when no better matching combination is generated. Finally, a unique binding beat point is determined for each decision quantum package. This two-way bidding mechanism ensures that high-potential decision quantum packages can be matched with beat points with optimal suitability, and also ensures that the carrying capacity of beat points is used reasonably, avoiding overload or waste of resources for a single beat point.

[0090] In a preferred embodiment of the present invention, the reverse simulation module is further configured to perform the following steps:

[0091] Step 41: Generate a characteristic perturbation wave for each bound beat point decision quantum packet. The initial amplitude of the characteristic perturbation wave is determined by the influence potential amplitude of the decision quantum packet, and the wavelength is determined by its action time span. The wave propagates attenuating along the time axis with the bound beat point of the decision quantum packet as the emission origin. The specific operation is as follows:

[0092] To quantify the perturbation effect of each decision quantum packet on the process energy density spectrum, characteristic perturbation waves need to be generated for the decision quantum packets with bound beat points. The purpose is to transform the abstract effect of the decision quantum packets into a propagable and superimposed physical wave model, providing concrete basic data for subsequent comprehensive perturbation analysis. The parameters of the characteristic perturbation waves are determined by the inherent properties of the decision quantum packets: the initial amplitude is directly related to the influence potential amplitude of the decision quantum packet; the larger the influence potential amplitude, the stronger the ability of the quantum packet to change the process state, and the larger the corresponding initial amplitude, ensuring that the wave intensity matches the command's ability; the wavelength is determined by the action time span of the decision quantum packet; the longer the action span, the longer the duration of the command's influence on the process, and the longer the corresponding wavelength, which conforms to the positive correlation between time span and wavelength in wave propagation. At the same time, the characteristic perturbation waves propagate bidirectionally along the time axis with the binding beat point of the decision quantum packet as the origin and gradually attenuate. This is because the influence of the decision command gradually dissipates with the distance from the origin in the time dimension, which conforms to the attenuation law of perturbation propagation in the process system. Based on the above characteristics, the mathematical expression of the characteristic perturbation waves is: ;

[0093] Based on the physical nature of wave propagation, this formula first simulates the periodicity of waves using a sine function, whose angular frequency is determined by the span of action. The decision is made to ensure that the wavelength matches the action time span; then the potential amplitude of the decision quantum packet's influence is considered. As the initial amplitude, the initial intensity of the fluctuation is quantified; finally, the decay effect of the disturbance over time is simulated through an exponential decay term, and the decay coefficient is... Controlling the decay rate, where, For the first The characteristic perturbation wave amplitude of a decision quantum packet at time t For the first The binding beat point moment of a decision quantum package; The disturbance attenuation coefficient is preset according to the response characteristics of the process system, with a value range of [0.1, 0.5]; t is any time. Through this generation process, the effect of each decision quantum packet is transformed into a characteristic disturbance wave with clear physical meaning.

[0094] Step 42: On the reverse time axis starting from the end of the sequence, superimpose the characteristic perturbation waves of all decision quantum packets, and determine their interference effect based on the phase of each characteristic perturbation wave to generate an interference superposition waveform with a comprehensive perturbation amplitude. The specific operation is as follows:

[0095] To comprehensively evaluate the overall perturbation effect of all decision quantum packets on the process energy density spectrum, it is necessary to superimpose all characteristic perturbation waves on the reverse time axis and analyze their interference effect. The aim is to obtain the comprehensive perturbation amplitude that reflects the synergistic effect of multiple instructions, avoiding the one-sidedness caused by single perturbation analysis. The choice of the reverse time axis stems from the execution logic of the decision quantum packet sequence: calculating backwards from the end of the sequence prioritizes the impact of subsequent instructions on earlier instructions, conforming to the simulation logic of aftereffect prioritization, ensuring that the superposition result can truly reflect the temporal correlation effect of instruction execution. During the superposition process, it is crucial to focus on the phase relationship of each characteristic perturbation wave to determine the interference effect: when the phase difference between two perturbation waves at the same moment is 0 or an integer multiple of 2π, constructive interference occurs, increasing the comprehensive perturbation amplitude; when the phase difference is an odd multiple of π, destructive interference occurs, decreasing the comprehensive perturbation amplitude; other phase differences correspond to different degrees of partial interference. Based on the wave superposition principle and the characteristics of the reverse time axis, the expression for calculating the comprehensive perturbation amplitude is: ;

[0096] First, the amplitude of each characteristic disturbance wave is phase corrected using a cosine term. Quantify the effective contribution of the wave at the current moment, phase angle The amplitude is determined by both the wave propagation time and the initial phase; then, the amplitudes of all corrected disturbance waves are summed to obtain the comprehensive disturbance amplitude, which fully integrates the synergistic effects of various commands. In the formula, Let n be the magnitude of the combined perturbation at time t, and n be the total number of decision quantum packets. , For the first The initial phase of each characteristic perturbation wave is determined by the adjustment dimension characteristics of the decision quantum packet (with a value range of [0, 2π)). Through this superposition process, the generated interferometric superposition waveform can accurately reflect the comprehensive perturbation intensity of all decision quantum packets at different times.

[0097] Step 43: The combined disturbance amplitude of the interferometric superimposed waveforms is mapped to a modified risk field for the process energy density spectrum at the corresponding cycle point through a risk transfer function. The modified risk field includes the expected amount of spectral value modification and the uncertainty range. The specific operation is as follows:

[0098] The risk transfer function is constructed based on statistical analysis of a large amount of historical process data. By sorting out the correspondence between the comprehensive disturbance amplitude and the actual change in the process energy density spectrum, a nonlinear mapping relationship adapted to the specific process is obtained, ensuring that the function can truly reproduce the mechanism by which the disturbance affects the spectrum value. The first dimension of modifying the risk field is the expected amount of spectrum value modification. Its calculation is achieved by transforming the comprehensive disturbance amplitude into the most likely spectrum value change value through the risk transfer function. For example, when the comprehensive disturbance amplitude is 0.6, the expected amount is obtained through function mapping. The first dimension excludes random factors and focuses on the core correlation between disturbances and spectral value changes. The second dimension is the uncertainty range of spectral value modification, which is calculated based on the dispersion of spectral value changes under similar disturbances in historical data. For example, when the uncertainty coefficient σ=0.13, the uncertainty range corresponding to the comprehensive disturbance amplitude of 0.6 is... This range clarifies the range in which the actual spectral value may deviate from the expected amount, fully considering uncontrollable factors such as random fluctuations in the process system, equipment response errors, and differences in material properties, ensuring that the risk assessment includes both core trends and potential extreme situations.

[0099] Step 44: The modified risk field is used as a potential disturbance source and substituted into the beat characteristic field model for prospective simulation. The stability factor of the relevant beat points is re-estimated, and it is verified whether the re-estimated stability factor is lower than the set threshold. The specific operation is as follows:

[0100] The simulation process uses the modified risk field generated in step 43 as a potential disturbance source, fully integrating it into the previously constructed beat characteristic field model. The simulation scope covers all production beat points related to the current decision quantum packet sequence, including each bound beat point and its adjacent related beat points, avoiding omission of any process steps that may be affected by the disturbance. The revaluation of the stability factor follows the dynamic evolution logic of the beat characteristic field. Based on the initial stability factor without considering the disturbance, it combines the expected amount and uncertainty range of the spectral value modification, while incorporating the disturbance suppression effect of the inertial damping factor and the downstream constraint effect of the pressure transmission coefficient, to calculate the new stability factor for each relevant beat point. For example, if the initial stability factor is 0.8, the expected amount of the spectral value modification... After correction by inertial damping and pressure transmission, the re-estimated stability factor is 0.65. During the simulation, the uncertainty range needs to be transformed into a random disturbance term. By introducing random variables that conform to the fluctuation characteristics of the process system, the influence of uncontrollable factors is simulated to ensure that the re-estimated results are more in line with reality. After the simulation is completed, all re-estimated stability factors are checked one by one. If the stability factor of any tick point is lower than the set threshold, it is determined that the decision quantum packet sequence has stability risk and needs to return to the previous steps for adjustment; if the stability factor of all tick points is higher than or equal to the threshold, the verification is passed.

[0101] In a preferred embodiment of the present invention, the proxy game module is further configured to perform the following steps:

[0102] Step 51: Calculate the intra-sequence conflict coefficients between the validated decision quantum packets. The conflict coefficients are determined based on the temporal proximity of their binding beat points and the interference degree of the effect prediction of the reverse perturbation simulation. Also, calculate the migration intention for each decision quantum packet. The migration intention is positively correlated with the sum of its conflict coefficients and negatively correlated with the stability factor margin of its current binding beat point. The specific operations are as follows:

[0103] First, the intra-sequence conflict coefficient between the validated decision quantum packets needs to be calculated. This calculation focuses on two key dimensions: First, the temporal proximity of the binding beat points. The closer the binding beat points of two decision quantum packets are on the time axis, the more easily the perturbations generated during instruction execution are superimposed, resulting in a higher conflict risk. Temporal proximity is usually obtained by normalizing the time difference between the two beat points; the smaller the time difference, the larger the corresponding proximity quantization value. Second, the effect interference of the reverse perturbation simulation prediction. This indicator is obtained by analyzing the interference effect after the superposition of characteristic perturbation waves from two decision quantum packets. If the combined perturbation amplitude exceeds the reasonable range of a single quantum packet perturbation, or significantly increases the risk of spectral value modification at local beat points, it indicates a high effect interference. The conflict coefficient is a weighted combination of these two quantization values. The weights are set according to the process's sensitivity to temporal and effect conflicts. For example, in a certain precision process, the weight of temporal proximity is 0.6, and the weight of effect interference is 0. 4. The final conflict coefficient ranges from 0 to 1. A larger value indicates a more severe conflict. Subsequently, migration intention is calculated for each decision quantum bag. The logic of migration intention is to balance conflict avoidance with the adaptability of the current binding beat point: it is positively correlated with the sum of the conflict coefficients of the quantum bag. The larger the sum of the conflict coefficients, the more conflicts the quantum bag faces in the current sequence, and the more inclined it is to adjust the binding beat point to avoid conflict. At the same time, it is negatively correlated with the stability factor margin of the current binding beat point. The stability factor margin is the difference between the stability factor of the current binding beat point and the set threshold. The larger the margin, the higher the adaptability and security of the current binding relationship, and the lower the necessity of migration. For example, if the sum of the conflict coefficients of a decision quantum bag is 0.7 and the stability factor margin of the current binding beat point is 0.3, its migration intention will be significantly higher than that of a quantum bag with a sum of conflict coefficients of 0.3 and a stability factor margin of 0.7. Through this quantification method, it is ensured that the migration intention can truly reflect the adjustment needs of the quantum bag.

[0104] Step 52: Based on the migration intention and the potential magnitude of the decision quantum package's influence, calculate the dynamic bargaining power of each decision agent, and assign game strategies to each decision agent based on this power. The specific operations are as follows:

[0105] The calculation of dynamic bargaining power integrates two key parameters: the migration willingness of the decision quantum package and the magnitude of its influence potential. A higher migration willingness indicates a stronger motivation for the decision agent to adjust the binding tick point, making it more inclined to proactively propose changes in the game, and exhibiting higher initiative and persistence in bargaining. A higher influence potential magnitude indicates a greater contribution of the decision quantum package to process optimization, making its instruction execution effect more critical to overall production quality, and giving it higher priority in the game. The quantification process of dynamic bargaining power typically involves a weighted fusion of these two parameters. For example, in a certain process, the weight of migration willingness is 0.4, and the weight of influence potential magnitude is 0.6. By adding the quantified values ​​of the two parameters in this proportion, the final dynamic bargaining power value is obtained, ranging from 0 to 1. A larger value indicates stronger bargaining power. Based on the differences in dynamic bargaining power, game strategies are assigned to each decision agent: decision agents with high bargaining power, such as those with a dynamic bargaining power value greater than 0.7, will gain priority in proposing and vetoing, and can prioritize selecting more suitable tick points and addressing incompatible ones. The strategy of rejecting reasonable proposals is to maximize the effectiveness of its own instructions. Decision-making agents with medium bargaining power, such as those with a dynamic bargaining power value between 0.4 and 0.7, adopt a compromise strategy, maintaining their own needs while making appropriate concessions to facilitate negotiation. Decision-making agents with low bargaining power, such as those with a dynamic bargaining power value less than 0.4, adopt an adaptation strategy, prioritizing the acceptance of reasonable proposals from agents with high bargaining power, and only raising adjustment requests when the execution of their own instructions is severely affected. This differentiated strategy ensures that the game process is orderly and efficient.

[0106] Step 53: Within the interval of tick points where the stability factor exceeds the set threshold, perform multiple rounds of negotiation according to the game strategy. For each round of negotiation, the proposed tick point change is verified by calling the reverse perturbation simulation. The verified changes are recorded as temporary binding protocols, and an optimization decision sequence is formed based on the final set of protocols. The specific operations are as follows:

[0107] The negotiation process is confined to a tick point range where the stability factor exceeds a set threshold. This range provides a safe adjustment space for the decision quantum package, preventing process fluctuations caused by binding unqualified tick points. The negotiation proceeds in an orderly manner according to the assigned game strategy. In the first round of negotiation, decision agents with high bargaining power propose tick point changes first, selecting the more suitable target tick point based on their own dynamic affinity ranking. Decision agents with medium and low bargaining power then propose their own changes or respond to others' proposals in turn. All tick point change proposals generated in each round of negotiation must be locally verified using reverse perturbation simulation. Verification involves simulating the change in the process energy density spectrum of the relevant tick points after the change proposal is executed, and re-evaluating the stability factors of these tick points. The relevant tick points include the changed bound tick points and their preceding and following related tick points. If the stability factors of all re-evaluated relevant tick points are not lower than the set threshold, the proposal is deemed valid and recorded as a temporary binding agreement. If any stability factor is lower than the threshold, the proposal is rejected, and the proposing party must readjust the target tick point and resubmit. Multiple rounds of negotiation continue, with each round optimizing the proposed content based on the previous round's temporary agreement. Agents with high bargaining power can adjust their proposals based on the verification results to strive for better adaptability, while agents with low bargaining power can make appropriate compromises within a safe range to advance the negotiation process, until all decision agents reach a satisfactory temporary agreement and no better proposed changes are generated. Finally, all verified temporary binding agreements are integrated to form an optimized decision sequence. This sequence resolves the internal conflicts in the initial binding and ensures the execution security and adaptability of each decision quantum package.

[0108] In a preferred embodiment of the present invention, the beat synchronization module is further configured to perform the following steps:

[0109] Step 61: For each decision quantum packet in the optimization decision sequence, construct a dynamic countdown bridge. The bridge uses its bound tick point as the theoretical zeroing time and dynamically adjusts its elapsed speed according to the deviation rate between the monitored actual production tick rate and the theoretical tick rate. At the same time, it calibrates the execution critical window around the zeroing point. The specific operations are as follows:

[0110] For each decision quantum packet in the optimization decision sequence, a dynamic countdown bridge needs to be built independently to ensure that the execution timing of each instruction can be adapted to the fluctuations in the production rhythm. The time base of the bridge is set to the binding tick point of that decision quantum packet. This tick point serves as the theoretical zeroing point of the bridge, clarifying the ideal time node when the instruction should be executed. To cope with the fluctuations in the tick rate in actual production, the bridge needs to monitor the difference between the actual production tick rate and the theoretical tick rate in real time and calculate the offset rate. The offset rate is obtained by dividing the difference between the actual tick rate and the theoretical tick rate by the theoretical tick rate. If the actual tick rate is higher than the theoretical value, the offset rate is positive; otherwise, it is negative. Based on this offset rate, the bridge will dynamically adjust its elapsed speed: when the offset rate is positive, it indicates that... If actual production is too fast, the bridge should appropriately slow down the ebb rate to avoid premature instruction execution. When the offset rate is negative, it indicates that actual production is too slow, and the bridge should appropriately speed up the ebb rate to prevent instruction execution from lagging. This dynamic adjustment ensures that the bridge's zeroing time is always consistent with the arrival time of the bound cycle point in actual production. At the same time, the bridge needs to calibrate the execution critical window around the zeroing point. The length of this window is determined based on the response time of the process equipment and the preparation time required for instruction execution. For example, if the equipment response time in a certain process is 1 cycle unit and the preparation time is 1 cycle unit, then the execution critical window is set from 1 cycle unit before the zeroing point to 1 cycle unit after the zeroing point, with a total length of 2 cycle units, to reserve sufficient time for subsequent instruction verification and compilation.

[0111] Step 62: When the dynamic countdown bridge enters its execution critical window, the real-time stability factor of the target cycle point and the current state deviation of the relevant process equipment are verified. Based on the verification results, the adjustment instructions contained in the decision quantum package are conditionally compiled, and the compiled instructions are issued when the bridge returns to zero. The specific operation is as follows:

[0112] When the dynamic countdown bridge enters the preset execution critical window, the system immediately initiates a dual-state verification process. The first verification is the real-time stability factor of the target beat point. This is calculated by real-time acquisition of the current process energy density spectrum data and the beat characteristic field model, determining whether the real-time stability factor is still higher than the set threshold. If the real-time stability factor is lower than the threshold, it indicates that the current process state does not meet the command execution conditions and requires subsequent adjustment. The second verification is the current state deviation of the relevant process equipment. The actual and target values ​​of the key operating parameters of the equipment are collected, and the deviation is calculated. For example, if the target temperature of a certain equipment is 180℃ and the current actual temperature is 78℃, the state deviation is 2℃. This clarifies whether the current operating state of the equipment is suitable for the adjustment command. Based on the results of the dual verification, the system conditionally compiles the adjustment command contained in the decision quantum package: if the real-time stability factor meets the threshold requirement and the equipment state deviation is within the allowable range (e.g., the deviation is less than 3℃), the command is compiled according to the original optimized parameters to ensure the optimization effect of the command; if the stability factor margin is sufficient but the equipment has a slight deviation, the parameters of the adjustment command can be slightly modified, for example, by multiplying the original adjustment range by a correction coefficient. , A value of 0.9 is used to reduce the impact of instructions on the equipment. If the stability factor is close to the threshold or the equipment deviation is large, the compilation strategy is temporarily adjusted to retain the core optimization target while reducing the adjustment intensity to avoid causing process fluctuations. When the dynamic countdown bridge reaches the zeroing time, the system immediately issues the compiled adjustment instructions. At this time, the actual production cycle happens to reach the binding cycle point of the decision quantum package, ensuring that the adjustment instructions are executed at the optimal time, which fully utilizes the optimization effect and ensures the stability of the process system.

[0113] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A workpiece quality online analysis and production decision optimization model, characterized in that, include: The spectrum generation module is used to generate a one-dimensional process energy density spectrum online. The process energy density spectrum is formed by dynamically assigning weights to multiple process control variables and performing nonlinear superposition. The factor construction module calculates the stability factor for each production cycle point based on the process energy density spectrum to construct the cycle characteristic field. The stability factor is determined based on the deviation of the current spectrum value from the target benchmark, the gradient of spectrum value change at adjacent cycle points, and the state logic values ​​of the downstream buffer and downstream process. The instruction splitting module is used to split the optimization decision instruction into multiple decision quantum packages and bind each decision quantum package to the future production tick point in the tick characteristic field where the stability factor exceeds a set threshold. The reverse simulation module is used to perform reverse perturbation simulation on the decision quantum packet sequence of the bound beat point. Starting from the end of the sequence, it calculates the amount of modification to the process energy density spectrum after each decision quantum packet takes effect, and checks whether it will cause the stability factor of any beat point to fall below the set threshold. The proxy game module is used to take the verified decision quantum package as a proxy and play a proxy game in the interval of the beat point where the stability factor exceeds the set threshold, so as to adjust its binding beat point and form an optimized sequence. The beat synchronization module is used to synchronize the actual production beat with the optimization sequence. When the beat reaches the binding beat point of any decision quantum package, it executes the adjustment instructions contained therein.

2. The online workpiece quality analysis and production decision optimization model according to claim 1, characterized in that, The spectrum generation module includes: For each process control variable, based on its current value relative to its own process constraint window and its acceleration of change, the instantaneous influence potential sequence changing with time is calculated; Based on the instantaneous influence potential sequence of each process control variable, a temporal entanglement matrix is ​​constructed, where each element of the matrix is ​​determined based on the waveform similarity and phase difference of each pair of instantaneous influence potential sequences; The instantaneous influence potential sequence of each process control variable is convolved with the dynamic convolution kernel constructed by the row vectors corresponding to the process control variables in the temporal entanglement matrix to generate the original density flow corresponding to each variable. At each moment, the dominant flow with the largest gradient among all the primary density flows is identified, and the energy of each primary density flow at the current moment is focused onto the characteristic frequency envelope of the dominant flow, which is then collapsed to generate a one-dimensional process energy density spectrum.

3. The online workpiece quality analysis and production decision optimization model according to claim 1 or 2, characterized in that, The factor construction module includes: Based on the difference between the current value and the target reference value of the process energy density spectrum, and combined with the local frequency characteristics of the process energy density spectrum, the influence cone projection vector is calculated and generated. Each component of the influence cone projection vector represents the expected influence amplitude of the current spectral value deviation on future continuous beat points. Based on the curvature and direction of change of the process energy density spectrum near the beat point, the inertial damping factor at the beat point is calculated.

4. The online workpiece quality analysis and production decision optimization model according to claim 3, characterized in that, The factor construction module is specifically used for: Based on the physical occupancy status of the downstream buffer and the ready signal of the downstream process, the pressure transmission coefficient is calculated using the pressure transmission model. Using the stability factor field distribution obtained from the previous iteration as the initial field distribution, the influence cone projection vector as the disturbance source, the inertial damping factor as the field medium absorption coefficient, and the pressure transmission coefficient as the boundary driving force, the stability factor at the current beat point is obtained and the stability factor field distribution is updated through field iterative equilibrium calculation.

5. The online workpiece quality analysis and production decision optimization model according to any one of claims 1, 2, 3 or 4, characterized in that, The instruction splitting module includes: The decision instructions are analyzed according to different adjustment dimensions. The potential impact amplitude is determined based on the adjustment magnitude and parameter sensitivity coefficient of each dimension. The action span is determined by combining the physical time window required for each dimension to take effect. In this way, multiple decision quantum packages with potential impact amplitude and action span attributes are generated. For each candidate beat point in the beat characteristic field whose stability factor exceeds a set threshold, its dynamic affinity to each decision quantum packet is calculated. The dynamic affinity is determined by the matching degree between the stability factor margin of the beat point and the influence potential amplitude of the decision quantum packet, as well as the degree of agreement between the action span of the decision quantum packet and the future process stability window of the beat point predicted based on the process energy density spectrum.

6. The online workpiece quality analysis and production decision optimization model according to claim 5, characterized in that, The instruction splitting module is specifically used to perform the following binding operations: Using the impact potential amplitude of each decision quantum package as the bidding power and the dynamic affinity of each tick point to each decision quantum package as the acceptance willingness, a multi-round two-way bidding matching is performed to determine the final tick point for binding each decision quantum package.

7. The online workpiece quality analysis and production decision optimization model according to claim 6, characterized in that, The reverse simulation module includes: For each bound beat point, a characteristic perturbation wave is generated for the decision quantum packet. The initial amplitude of the characteristic perturbation wave is determined by the influence potential amplitude of the decision quantum packet, the wavelength is determined by its action time span, and it propagates attenuated along the time axis with the bound beat point of the decision quantum packet as the emission origin. On the reverse time axis starting from the end of the sequence, the characteristic perturbation waves of all decision quantum packets are superimposed, and their interference effect is determined according to the phase of each characteristic perturbation wave, generating an interference superposition waveform with a comprehensive perturbation amplitude.

8. The online workpiece quality analysis and production decision optimization model according to claim 7, characterized in that, The reverse simulation module is also used for: The combined disturbance amplitude of the interferometric superimposed waveform is mapped to a modification risk field of the process energy density spectrum at the corresponding tick point through the risk transfer function. The modification risk field includes the expected amount and uncertainty range of the spectrum value modification. The modified risk field is used as a potential disturbance source and substituted into the beat characteristic field model for prospective simulation. The stability factor of the relevant beat point is re-estimated, and it is verified whether the re-estimated stability factor is lower than the set threshold.

9. The online workpiece quality analysis and production decision optimization model according to claim 6 or 8, characterized in that, The proxy game module includes: The intra-sequence conflict coefficients between each decision quantum bag that has been verified are calculated. The conflict coefficients are determined based on the temporal proximity of their binding beat points and the effect interference of the reverse perturbation simulation prediction. The migration intention is calculated for each decision quantum bag. The migration intention is positively correlated with the sum of its conflict coefficients and negatively correlated with the stability factor margin of its current binding beat point. Based on migration intention and the potential magnitude of the influence of the decision quantum package, calculate the dynamic bargaining power of each decision agent, and assign game strategies to each decision agent based on this power; Within the interval of beat points where the stability factor exceeds the set threshold, multiple rounds of negotiation are executed according to the game strategy. For each round of negotiation, the proposed beat point change is verified by calling the reverse perturbation simulation. The verified changes are recorded as temporary binding protocols, and an optimization decision sequence is formed based on the final set of protocols.

10. The online workpiece quality analysis and production decision optimization model according to claim 9, characterized in that, The beat synchronization module includes: For each decision quantum packet in the optimization decision sequence, a dynamic countdown bridge is constructed. The bridge takes its bound tick point as the theoretical zeroing time and dynamically adjusts its elapsed speed according to the deviation rate between the monitored actual production tick rate and the theoretical tick rate. At the same time, the execution critical window around the zeroing point is calibrated. When the dynamic countdown bridge enters its execution critical window, it verifies the real-time stability factor of the target tick point and the current state deviation of the relevant process equipment, and conditionally compiles the adjustment instructions contained in the decision quantum package based on the verification results, and issues the compiled instructions when the bridge returns to zero.

Images