Rubber cable production quality control system and method

Abstract

The invention belongs to the technical field of rubber cable production, and provides a rubber cable production quality control system and method. The method comprises the following steps: acquiring raw material formula data and raw material mixing data; predicting raw material mixing uniformity based on the raw material mixing data, and generating interference sequence data based on the raw material mixing uniformity; processing the raw material formula data and the interference sequence data by using a multi-parameter collaborative optimization model to obtain target control parameters; and controlling the rubber extrusion operation of the extrusion equipment according to the target control parameters. According to the method, target control parameters, namely the optimal extrusion speed, extrusion temperature and extrusion pressure, are obtained by using a multi-parameter collaborative optimization model; in addition, the influence of raw material mixing uniformity on extrusion control is also considered in the collaborative optimization process. Therefore, the scheme provided by the invention can effectively reduce the probability that the extruded rubber has a quality problem, so that the thickness of a cable conductor cladding layer is more accurate.

Description

Technical Field

[0001] This invention relates to the field of rubber cable production technology, and more specifically, to a quality control system and method for rubber cable production. Background Technology

[0002] In the production of rubber cables, quality control is a crucial step in ensuring stable and reliable product performance. Precise control of key parameters such as extrusion speed, temperature, and pressure directly affects the cable's dimensional accuracy, surface quality, and physical properties. However, due to the complexity and dynamic nature of the extrusion process, it is often difficult to achieve real-time, precise control of these parameters in existing technologies, leading to instability in the extrusion process and consequently impacting the quality of the final product.

[0003] Specifically, current extrusion speed control technologies typically rely on experience-based adjustments, lacking a dynamic feedback mechanism and making it difficult to adapt to changes in material properties or process conditions. Regarding temperature control, because rubber materials are sensitive to temperature, temperature fluctuations can easily lead to uneven material performance or extrusion defects. Pressure control, due to the complex flow field inside the extruder, makes high-precision real-time monitoring and adjustment difficult using traditional methods. These problems not only reduce production efficiency but also increase scrap rates and production costs.

[0004] Therefore, there is an urgent need for a new quality control method for rubber cable production to achieve precise control and stable output of key parameters such as extrusion speed, temperature, and pressure, thereby improving the product quality and production efficiency of rubber cables. Summary of the Invention

[0005] In response, the present invention provides a method, system, electronic device, computer storage medium, and computer program product for quality control in rubber cable production, to solve at least one of the above-mentioned technical problems.

[0006] In a first aspect, the present invention provides a method for quality control in the production of rubber cables, comprising the following steps: acquiring raw material formulation data and raw material mixing data; the raw material formulation data including the proportioning data of rubber and its additives, and the raw material mixing data including the mixing data of the batch of raw materials by a mixing equipment; predicting the raw material mixing uniformity based on the raw material mixing data, and generating interference sequence data based on the raw material mixing uniformity; processing the raw material formulation data and the interference sequence data using a multi-parameter collaborative optimization model to obtain target control parameters, including extrusion speed, extrusion temperature, and extrusion pressure; and controlling the rubber extrusion operation of the extrusion equipment according to the target control parameters.

[0007] In a second aspect, the present invention provides a quality control system for rubber cable production. The system includes an acquisition unit, an optimization and compensation unit, and a control unit. The acquisition unit is used to acquire raw material formulation data and raw material mixing data. The raw material formulation data includes the proportioning data of rubber and its additives, and the raw material mixing data includes the mixing data of the batch of raw materials by the mixing equipment. The optimization and compensation unit is used to predict the raw material mixing uniformity based on the raw material mixing data and to generate interference sequence data based on the raw material mixing uniformity. A multi-parameter collaborative optimization model is used to process the raw material formulation data and the interference sequence data to obtain target control parameters, including extrusion speed, extrusion temperature, and extrusion pressure. The control unit is used to control the rubber extrusion operation of the extrusion equipment according to the target control parameters.

[0008] In a third aspect, the present invention provides an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program, when executed by the processor, implements the method as described in any of the preceding claims.

[0009] In a fourth aspect, the present invention provides a computer storage medium storing a computer program that can be executed by a processor to implement the method as described in any of the preceding claims.

[0010] In a fifth aspect, the present invention provides a computer program product comprising a computer program executable by a processor to implement the method as described in any of the preceding claims.

[0011] The beneficial technical effects of this invention are as follows: For specific raw material formulation data, this invention obtains the target control parameters—namely, the optimal extrusion speed, extrusion temperature, and extrusion pressure—through a multi-parameter collaborative optimization model. Furthermore, the influence of raw material mixing uniformity on extrusion control is considered during the collaborative optimization process. Therefore, this invention effectively reduces the probability of quality problems in the extruded rubber, resulting in more precise thickness of the cable conductor sheath, improved surface quality, compliance with physical performance standards, and a reduced scrap rate. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1This is a flowchart illustrating a method for quality control in the production of rubber cables, as disclosed in an embodiment of the present invention.

[0014] Figure 2 This is a schematic diagram of the scenario architecture disclosed in an embodiment of the present invention.

[0015] Figure 3 This is a schematic diagram of the structure of a rubber cable production quality control system disclosed in an embodiment of the present invention. Detailed Implementation

[0016] The following specific embodiments illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] Furthermore, the technical features involved in the different embodiments of this application described below can be combined with each other as long as they do not conflict with each other.

[0018] like Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for quality control in the production of rubber cables, including the following steps: S10, obtaining raw material formula data and raw material mixing data; the raw material formula data includes the proportion data of rubber and its additives, and the raw material mixing data includes the mixing data of the mixing equipment for this batch of raw materials.

[0019] like Figure 2 As shown, the solution of the present invention is executed by a rubber cable production management system, which is connected to at least a mixing device, an extrusion device, and a management terminal. The mixing device is used to mix the rubber raw materials of this batch to achieve homogenization. The extrusion device adjusts the appropriate extrusion speed, extrusion temperature, and extrusion pressure to achieve stable extrusion of qualified rubber, thereby ensuring the quality of the rubber cable. The management terminal is used to receive raw material formula data input or selected by the management personnel.

[0020] First, the production management system receives raw material formula data input or selected by management personnel from the management terminal, including the proportions of rubber and its additives, such as 75% rubber, 15% flame retardant, 8% plasticizer, and 2% antioxidant. It is understood that the rubber raw materials in this invention can be natural rubber, irradiated rubber, nitrile rubber (NBR), styrene-butadiene rubber (SBR), chloroprene rubber (CR), silicone rubber (Q), chlorinated polyethylene rubber (CM or CPE), etc., without specific limitations.

[0021] In addition, the production management system also receives actual mixing operation data of the batch of rubber raw materials from the mixing equipment, i.e., raw material mixing data. For example, when using a twin-ribbon mixer to mix this batch of rubber raw materials, the mixing data is collected in real time by sensors built into the mixing equipment. The raw material mixing data may be as follows: the mixer stirring speed is 80 r / min, the mixing time is set to 30 min, and the temperature inside the drum is maintained at 60℃±5℃ during the mixing process.

[0022] S20, the raw material mixing uniformity is predicted based on the raw material mixing data, and interference sequence data is generated based on the raw material mixing uniformity; the raw material formulation data and the interference sequence data are processed using a multi-parameter collaborative optimization model to obtain target control parameters, including extrusion speed, extrusion temperature, and extrusion pressure.

[0023] The production management system employs appropriate algorithms or models to analyze the collected raw material mixing data. For example, a model trained on historical production data can be used to establish a correlation between factors such as mixing time, rotation speed, and temperature and the actual mixing uniformity. Based on a table showing the relationship between mixing uniformity and extrusion parameter interference, the production management system generates interference sequence data. For instance, for every 1% decrease in mixing uniformity, the extrusion speed needs to be reduced by 0.5 m / min, the extrusion temperature needs to be increased by 1°C, and the extrusion pressure needs to be increased by 0.2 MPa.

[0024] For example, the predicted uniformity of the raw material mixture for this batch is 82% (the standard for uniformity is set at ≥90%), indicating an uneven mixing problem. For instance, an 8% decrease in uniformity corresponds to a 4m / min decrease in extrusion speed interference, an 8℃ increase in extrusion temperature interference, and a 1.6MPa increase in extrusion pressure interference.

[0025] Raw material formulation data and interference sequence data are input into a multi-parameter collaborative optimization model. This model, constructed based on a genetic algorithm, comprehensively considers the rheological properties of rubber materials and the performance parameters of extrusion equipment, and iteratively calculates the target control parameters. The final target control parameters for this batch of production are, for example: extrusion speed 18 m / min (current value 22 m / min), extrusion temperature 185℃ (current value 177℃), and extrusion pressure 12.6 MPa (current value 11 MPa).

[0026] The general optimization process of a multi-parameter collaborative optimization model is illustrated below with an example: 1. Model construction and objective function definition: Construct a collaborative optimization framework including a raw material formulation sub-model, a disturbance response sub-model, and a process constraint layer. With the objectives of improving mixing uniformity, enhancing interference compensation, and satisfying process constraints, the loss function is defined as follows: ;in, The target control parameters to be optimized are (extrusion speed, extrusion temperature, and extrusion pressure).

[0027] To quantify the adaptation error between raw material formulation and parameters, based on formulation data... (e.g., component proportions, molecular weight distribution) and target parameters The matching degree is calculated using the mean squared error method: , For parameters The appropriate mixing ratio of the i-th raw material. This refers to the process tolerance of the raw material.

[0028] Characterizing the response loss of the perturbed sequence; based on perturbed data With target parameters The compensation effect is calculated using absolute value loss: , For target parameters The compensation amount for the j-th type of interference is as follows: (1) Speed ​​disturbance compensation amount (used to offset) ): ;in, It is the maximum speed adjustment.

[0029] (2) Temperature interference compensation amount (used to offset) ): ;in, The gas constant is... This is the initial reference temperature, i.e., the ideal reference temperature for the production process. The activation energy of the material. For ideal viscosity, This represents the actual viscosity after temperature disturbance.

[0030] (3) Pressure disturbance compensation amount (used to offset) ): ;in, This represents the maximum allowable adjustment range for pressure.

[0031] To mitigate process constraint losses, hard boundary constraints (such as speed) are set for the target parameters. (using a penalty function) .

[0032] The weights are denoted as and the sum is 1. Since equipment can be replaced or repaired, but product defects are unacceptable, and the probability of process constraints being triggered in normal production is very low (because they can be avoided through process design), and equipment is usually equipped with redundant safety designs (such as a forced shutdown if a pressure sensor exceeds its range), therefore, setting... With an emphasis on formula adaptation, The focus is on interference compensation.

[0033] 2. Raw material formulation feature processing: processing of raw material formulation data. Extracting multi-scale features: Component proportion features: Calculating the entropy value of the mass fraction of each raw material. , ( (total mass), characterizing the uniformity of the formulation.

[0034] Rheological property mapping: through capillary rheological models, such as power-law capillary rheological models. Convert the formulation data to equivalent viscosity Establish parameters related to the target The connection. Among them, Equivalent viscosity characterizes the ease or difficulty of material flow; It is the consistency coefficient, which is related to the material's inherent properties (such as the rigidity of rubber molecular chains and the dispersibility of fillers), and reflects the viscous characteristics of the material at low shear rates; Shear rate describes the intensity of shearing action experienced by a material during processing (such as extruder screw rotation or mixing). It reflects how quickly the material is sheared and deformed. Its value is directly related to equipment operating parameters (such as extruder speed). Higher speeds and more specialized shear component structures result in higher shear rates. Generally, the larger; Rheological index, used to distinguish the rheological type of material ( It is a Newtonian fluid. As a non-Newtonian fluid, rubber materials are typically (This exhibits pseudoplasticity).

[0035] In the rubber cable extrusion process, shear rate The shear rate is mainly determined by the extruder screw speed N (unit: r / min) and screw structure (screw groove depth h, screw diameter D, etc., which are inherent parameters of the equipment). The shear rate is obtained through theoretical derivation or experimental calibration. The formula relating screw speed N to screw rotation speed (taking a common single-screw extruder as an example): .

[0036] Mapped features Input the formula sub-model to obtain the initial parameters. .

[0037] 3. Dynamic compensation for interference sequences: This involves compensating for interference sequence data. Perform joint time-frequency domain analysis: Time-domain compensation: based on the attenuation / amplification characteristics of interference (e.g.) Predicting the future Step interference trend .

[0038] Frequency domain compensation: Extracting the dominant frequency of the interference sequence using Fast Fourier Transform (FFT). Matching the resonant frequency of the equipment To avoid resonance caused by parameter adjustments.

[0039] By combining trend prediction and frequency analysis, an interference compensation matrix is ​​constructed. ,in, . For the maximum adjustment of parameter j, the initial parameter is... Make compensation adjustments: .

[0040] 4. Collaborative Optimization and Constraint Satisfaction: An Adaptive Genetic Algorithm (AGA) is employed to... To achieve the goal within the process constraint range Optimization, iteration until continuous The rate of change of the optimal solution < (like Output target parameters .

[0041] 5. Parameter verification feedback: Verified through production simulation. (e.g., using a CFD-DEM coupled model), for example, verifying mixing uniformity and product mechanical properties (tensile strength, elongation at break, etc.); if not satisfied, adjust the interference compensation weights. Optimize again until the performance meets the requirements.

[0042] S30, control the rubber extrusion operation of the extrusion equipment according to the target control parameters.

[0043] The target control parameters calculated in step S20 are sent to the extrusion equipment control system to perform corresponding synchronous control on the extrusion equipment drive motor, heating device, and pressure regulating device.

[0044] For example, the extrusion equipment drive motor adjusts its speed according to the target extrusion speed of 18m / min, the heating device stabilizes the temperature at 185℃ through a PID control algorithm, and the pressure regulating device monitors in real time and maintains the extrusion pressure at 12.6MPa.

[0045] In addition, throughout the extrusion process, the extrusion equipment control system needs to continuously collect actual working parameters and compare them with target control parameters. If the deviation exceeds the set threshold (speed ±0.5m / min, temperature ±2℃, pressure ±0.3MPa), the extrusion equipment control system will automatically make fine adjustments to ensure stable operation of the extrusion process.

[0046] In this invention, for specific raw material formulation data, the target control parameters, namely the optimal extrusion speed, extrusion temperature, and extrusion pressure, are obtained by using a multi-parameter collaborative optimization model. Furthermore, the influence of raw material mixing uniformity on extrusion control is also considered during the collaborative optimization process. Therefore, this invention can effectively reduce the probability of quality problems in the extruded rubber, thereby resulting in more precise cable conductor sheath thickness, improved surface quality, compliance with physical performance standards, and a reduced scrap rate.

[0047] As an example, generating interference sequence data based on the raw material mixing uniformity includes: determining whether the raw material mixing uniformity is higher than a preset threshold; if so, not generating interference sequence data based on the raw material mixing uniformity; if not, generating interference sequence data based on the raw material mixing uniformity; wherein the preset threshold is derived based on at least one of the following: rubber material characteristics, rubber cable performance requirements, and extrusion equipment performance.

[0048] This invention designs the aforementioned logic for generating interference sequence data to reduce the increased computational burden on generating target control parameters due to unnecessary interference sequence data, and to reduce the probability of insufficient confidence in the generated target control parameters caused by erroneous interference. Specifically: when the raw material mixing uniformity is higher than a preset threshold, it indicates good raw material mixing uniformity, and it is unlikely to significantly interfere with the subsequent extrusion process; therefore, there is no need to generate interference sequence data to adjust the extrusion parameters. If the raw material mixing uniformity is lower than the preset threshold, it indicates poor mixing effect, which may affect extrusion stability and product quality. In this case, interference sequence data needs to be generated based on the raw material mixing uniformity to correct parameters such as extrusion speed, extrusion temperature, and extrusion pressure to ensure the production quality of rubber cables.

[0049] The preset threshold can be determined based on at least one of the following factors: Rubber material characteristics: Different rubbers and additives have different rheological properties and compatibility, resulting in different requirements for mixing uniformity. For example, high-viscosity rubbers or combinations of additives with poor compatibility require higher mixing uniformity to ensure stable material flow during extrusion. Therefore, the preset threshold will be increased accordingly based on the rubber material characteristics.

[0050] Performance requirements for rubber cables: High-voltage and special rubber cables have stringent requirements for insulation and mechanical strength, requiring higher mixing uniformity to ensure quality. Their preset thresholds are higher than those for low-voltage ordinary cables. Extrusion equipment performance: The screw structure, heating capacity, and pressure control precision of the extrusion equipment affect its adaptability to raw material mixing uniformity. Advanced extrusion equipment can be compatible with raw materials with slightly lower mixing uniformity to a certain extent, and its preset threshold can be appropriately lowered.

[0051] Understandably, when choosing to determine the preset threshold based on the above three factors, the Analytic Hierarchy Process (AHP) can be used to first determine the importance weight of each factor, then fuzzy logic reasoning can be used to calculate the score corresponding to each factor, and finally the importance weight can be used to integrate the scores into the preset threshold.

[0052] The process of using fuzzy logic reasoning to calculate the scores of each factor is roughly as follows: 1. Input variables: Rubber material properties: viscosity (low / medium / high), compatibility (good / medium / poor).

[0053] Cable performance requirements: voltage rating (low / medium / high), mechanical strength (ordinary / reinforced / special).

[0054] Extrusion equipment performance: screw length-to-diameter ratio (large / medium / small), control precision (high / medium / low).

[0055] 2. Examples of fuzzy rules: IF high material viscosity AND high cable voltage rating THEN threshold = high; IF high equipment precision AND good material compatibility THEN threshold = medium.

[0056] 3. Output: Preset threshold T (0-100% range).

[0057] Alternatively, neural network models (NN), genetic algorithms, etc., can also be used, but details will not be elaborated here.

[0058] As an example, the step of predicting the raw material mixing uniformity based on the raw material mixing data and generating interference sequence data based on the raw material mixing uniformity includes: extracting multimodal features, including physical features, dynamic features, and temporal features, from the raw material mixing data; using an ensemble prediction model based on a stacked ensemble framework to predict the multimodal features to obtain the raw material mixing uniformity; wherein the feature weights of each base model in the ensemble prediction model are dynamically adjusted based on the mixing duration; inputting the raw material mixing uniformity and the real-time observation sequence into a hidden Markov model, using the Viterbi algorithm to solve for the optimal state path, and calculating the state prediction distribution for the next n steps; wherein the hidden Markov model defines multiple stages; and using a stage-parameter mapping function to generate a dynamically adjusted sequence of each interference quantity changing over time, i.e., the interference sequence data.

[0059] First, multimodal features, including physical features, dynamic features, and temporal features, are extracted from the raw material mixing data.

[0060] Physical characteristics: Directly measurable static parameters (such as mixing time, rotational speed, and temperature), reflecting the fundamental conditions of the mixing process; Dynamic characteristics: Parameter fluctuation characteristics (such as torque fluctuation coefficient). , These represent the maximum, minimum, and average torque values ​​during the mixing process, capturing the unsteady changes during mixing.

[0061] Time-series characteristics: The mixing process is divided into n time windows, and the rate of change of parameters dX / dt within each window is extracted. X is, for example, the rotational speed, temperature, torque fluctuation coefficient, etc., which characterizes the dynamic trend of the mixing process.

[0062] The extracted multimodal features are processed using an ensemble prediction model within a stacking framework to obtain the raw material mixing uniformity. This ensemble prediction model incorporates multiple base models, such as XGBoost, Random Forest, and Support Vector Regression (SVR), leveraging the strengths of each model simultaneously. For example, XGBoost offers excellent nonlinear fitting, Random Forest provides robust noise reduction, and SVR offers high generalization ability, ensuring the accuracy of the obtained raw material mixing uniformity. It is understood that the raw material mixing uniformity is continuously monitored and predicted based on the mixing and stirring process of the mixing equipment.

[0063] The ensemble prediction model requires weighted integration of the prediction results from each base model to obtain the final prediction result. This invention sets the feature weights of each base model during the weighted integration process to be dynamically adjusted based on the mixing duration. Specifically, the importance of features is adjusted according to the mixing duration (e.g., emphasizing temporal features in the early stages and physical features in the later stages) to adapt to the time-varying characteristics of the mixing process. For example, when the mixing duration t < 15 min, the weights are assigned as follows: temporal features 0.4, physical features 0.3, and dynamic features 0.3; when the mixing duration t ≥ 15 min, the weights are adjusted to: temporal features 0.3, physical features 0.4, and dynamic features 0.3.

[0064] Then, a hidden Markov model is used to predict the fluctuation phase, as follows: 1. Model construction: State space: Define 4 hidden states (stable phase, initial fluctuation, significant fluctuation, violent fluctuation) to quantify the different degrees of uniformity fluctuation.

[0065] S1 (Stable Stage): Fluctuation range ≤ ±5%, corresponding to the stable state of extrusion parameters.

[0066] S2 (Initial Fluctuation): Fluctuation range of 5%-10%, the uniformity of raw material mixing begins to decline but does not significantly affect extrusion.

[0067] S3 (Significant fluctuation): Fluctuation range of 10%-15%, which may lead to extrusion pressure pulsation and surface quality defects.

[0068] S4 (Severe fluctuation): Fluctuation amplitude >15%, which is very likely to cause equipment vibration and excessive deviation of cable outer diameter.

[0069] Observation sequence: Selected extrusion pressure fluctuations (Reflecting changes in material flow resistance), extrusion temperature gradient (Characterizing plasticization uniformity), extrusion motor current change rate (Indirectly reflecting material viscosity fluctuations) as an observed variable.

[0070] Pre-processed using the Baum-Welch algorithm based on historical data ( , For the first The observation sequence at time, For the first The hidden state at time step (the transition probability matrix A, such as the probability of transitioning from steady to fluctuating), the observation probability matrix B, such as the probability of pressure change in a fluctuating state, and the initial state distribution are estimated. This completes the parameter training process. The convergence condition is the log-likelihood function. The rate of change is <0.01%. The model parameters are A, B, composition.

[0071] Next, mix the raw materials evenly. and real-time observation sequence Input the above Hidden Markov Model and use the Viterbi algorithm to solve for the optimal state path: ;in, for Moment State The most probable path, Let be the state transition probability. This represents the observation probability.

[0072] Record the optimal state at each step to generate a fluctuation stage estimate for the "current moment". This refers to the "stable phase," "initial fluctuation," "significant fluctuation," or "violent fluctuation." Understandably, this refers to the estimation of the fluctuation phase at the "current moment." In subsequent algorithms, it is used as an anchor point for the current state, that is, to predict the future state based on the current state.

[0073] The probability distribution of future states is calculated by extending the algorithm forward: ; Calculate the future recursively The state probabilities of each step are analyzed, with a focus on the occurrence probabilities of S3 and S4.

[0074] Design and use a phase-parameter mapping function. ,in, ;in, For speed disturbance quantity, For temperature disturbance quantity, This refers to the pressure disturbance quantity; These are the current extrusion speed, extrusion temperature, and extrusion pressure; the speed mapping coefficient. Temperature mapping coefficient Pressure mapping coefficient Velocity-probability decay coefficient Pressure-probability amplification factor . The future time calculated by the forward algorithm is in or The probability of a fluctuation phase.

[0075] Generate time-domain interference sequences: Among them, the attenuation coefficient Time step , This is the prediction step size for the multi-parameter collaborative optimization model.

[0076] As an example, the method of using the Viterbi algorithm to solve for the optimal state path includes: dividing the state transition grid of the Hidden Markov Model into m parallel processing units according to the time dimension, with each unit independently calculating the local optimal probability value of each state at the current time step; using a pruning strategy to retain the top k% of the state paths with the highest probability values; and using a bidirectional backtracking mechanism to perform bidirectional backtracking on the pruned state paths, completing the integration of the optimal paths at the intermediate intersection point, thereby obtaining the optimal state path.

[0077] In the production process of rubber cables, fluctuations caused by uneven mixing of raw materials need to be compensated for in real time. The time complexity of the traditional Viterbi algorithm is O(T×N). 2 (T is the number of time steps, and N is the number of states). When N is large (e.g., multi-level states of quality fluctuations) or T is long (e.g., continuous production processes), the calculation delay is significant (approximately 50-100ms in actual measurements), which may cause the compensation action to lag behind the occurrence of the fluctuation, affecting product quality. To address this, this invention optimizes and improves the Viterbi algorithm to enhance the response speed and meet the requirements of industrial real-time control (typically <20ms).

[0078] First, the state transition grid of the Hidden Markov Model is divided into m processing units according to the time dimension (e.g., when m=4, the time steps of T=100 are divided into [1-25], [26-50], [51-75], [76-100]), and each unit is processed in parallel by an independent computing core (e.g., GPU thread block).

[0079] Specifically: Each unit synchronously calculates the probabilities of each state at the current time step t: ;in, For the first The state of each unit at time t The local optimal probability.

[0080] In this way, the computation time complexity is reduced from O(T) to O(T / m), and when m=4, the theoretical speedup is 4 times.

[0081] Then, after calculating the local optimal probability at each time step t, all states are immediately sorted by probability value, and the top k% (e.g., k=20) of state paths are retained. By eliminating low-probability branches, the number of state paths that need to be processed in subsequent calculations can be significantly reduced, thus lowering the time and space complexity of the algorithm.

[0082] The bidirectional backtracking mechanism is built upon the aforementioned pruning strategy. The significantly reduced number of paths retained after pruning provides the prerequisite for efficient bidirectional backtracking. Path backtracking is performed simultaneously from both the start and end points of the retained state paths, merging the paths at intermediate intersection points. By performing bidirectional backtracking on the limited number of pruned state paths, not only can the merging of the two ends of the paths be completed quickly, but also ineffective computations on a large number of redundant paths are avoided.

[0083] Specifically: the time series is divided into the first half [1, T / 2] and the second half [T / 2+1, T], and backtracking is started from t=1 and t=T respectively.

[0084] Forward backtracking: Backtracking from t=T along the retained high-probability path to t=T / 2, generating path segments. .

[0085] Backtracking: Starting from t=1, trace forward along the preserved path to t=T / 2, generating path segments. .

[0086] Path integration: Verify the consistency of the states of the two segments at t=T / 2. If they are inconsistent, select the combination with the largest probability product.

[0087] In this way, the backtracking time complexity is reduced from O(T) to O(T / 2), which further improves the solution speed of the optimal state path based on the pruning strategy, thereby increasing the generation rate of interference sequence data.

[0088] As an example, the method of retaining the top k% of state paths with the probability value using a pruning strategy includes: calculating a fluctuation coefficient based on the raw material mixing uniformity and a preset mixing degree threshold, and calculating the entropy value of the state transition probability distribution at the current time step; calculating a k value based on the fluctuation coefficient and the entropy value, and retaining the top k% of state paths with the probability value using a pruning strategy.

[0089] In rubber cable production, if the k value is too small, excessive pruning will lead to the loss of potential optimal paths, resulting in prediction errors; if the k value is too large, it will be impossible to effectively reduce the computational load and meet real-time requirements. To balance computational efficiency and the prediction accuracy of state paths, this embodiment further determines the k value by integrating two major factors: the raw material mixing state and the uncertainty of model prediction, thereby achieving dynamic optimization of the pruning strategy. Specifically, the fluctuation coefficient reflecting the mixing quality risk is calculated: the uniformity of raw material mixing directly affects cable quality. By calculating the deviation ratio (fluctuation coefficient) between the current mixing degree prediction value and the preset threshold, the quality risk is quantified. The larger the fluctuation coefficient, the worse the raw material mixing uniformity, and the higher the quality risk. In this case, more state paths should be retained to avoid missing the optimal solution, corresponding to a larger k value; conversely, the number of paths retained can be reduced.

[0090] Next, the state transition entropy, which reflects the uncertainty of prediction, is calculated: the state transition entropy measures the randomness of state transitions in a hidden Markov model. The higher the entropy value, the greater the uncertainty of the model's prediction. In this case, conservative pruning is required to retain more state paths to ensure accuracy, corresponding to a larger k value; the lower the entropy value, the more aggressive pruning strategy can be adopted.

[0091] Next, the fluctuation coefficient and state transition entropy are divided into different intervals, each interval corresponding to a basic value of k. For example, when the fluctuation coefficient is less than 0.1, the mixed state is considered stable, and a basic value of k is taken. When the state transition entropy is greater than 2.0, the prediction uncertainty is considered high, and a value of 2.0 is taken. .

[0092] The weights of the fluctuation coefficient and state transition entropy are dynamically adjusted according to the production stage. In the initial stage of mixing, the raw materials have large differences, and the fluctuation of the raw material mixing uniformity is the main risk. Therefore, a higher weight is given to the fluctuation coefficient (e.g., In the later stages of mixing, the uncertainty of the extrusion equipment response becomes more prominent, thus reducing the weight of the fluctuation coefficient (e.g.) ), through weighted formula Calculate the final value of k.

[0093] Understandably, safety thresholds and verification mechanisms can also be set to prevent excessive pruning. For example, a probability threshold can be set for the paths to be retained. If the probability of a path is too low but still higher than the threshold by a certain percentage (e.g., 0.8 times), it will be forcibly retained. At the same time, the probability change rate of the optimal state path before and after pruning can be compared periodically. If the probability change rate exceeds the threshold (e.g., 5%), the k value will be automatically adjusted to ensure the reliability of the pruning strategy.

[0094] like Figure 3 As shown in the figure, this embodiment of the invention also provides a rubber cable production quality control system 100. The system 100 includes an acquisition unit 1001, an optimization compensation unit 1002, and a control unit 1003. The acquisition unit 1001 is used to acquire raw material formula data and raw material mixing data. The raw material formula data includes the proportion data of rubber and its additives, and the raw material mixing data includes the mixing data of the mixing equipment for this batch of raw materials.

[0095] The optimization compensation unit 1002 is used to predict the raw material mixing uniformity based on the raw material mixing data, generate interference sequence data based on the raw material mixing uniformity, and process the raw material formulation data and the interference sequence data using a multi-parameter collaborative optimization model to obtain target control parameters, including extrusion speed, extrusion temperature, and extrusion pressure.

[0096] The control unit 1003 is used to control the rubber extrusion operation of the extrusion equipment according to the target control parameters.

[0097] As an example, the optimization compensation unit 1002 specifically: determines whether the uniformity of the raw material mixing is higher than a preset threshold; if so, it does not generate interference sequence data based on the uniformity of the raw material mixing; if not, it generates interference sequence data based on the uniformity of the raw material mixing; wherein, the preset threshold is derived based on at least one of the characteristics of rubber materials, the performance requirements of rubber cables, and the performance of extrusion equipment.

[0098] As an example, the optimization compensation unit 1002 specifically: extracts multimodal features, including physical features, dynamic features, and temporal features, from the raw material mixing data; uses an ensemble prediction model based on a stacked ensemble framework to predict the multimodal features to obtain the raw material mixing uniformity; wherein, the feature weights of each base model in the ensemble prediction model are dynamically adjusted based on the mixing duration; inputs the raw material mixing uniformity and real-time observation sequence into a hidden Markov model, uses the Viterbi algorithm to solve for the optimal state path, and calculates the state prediction distribution for the next n steps; wherein, the hidden Markov model defines multiple stages; uses a stage-parameter mapping function to generate a dynamically adjusted sequence of each interference quantity changing over time, i.e., the interference sequence data.

[0099] As an example, the optimization compensation unit 1002 specifically: divides the state transition grid of the Hidden Markov Model into m parallel processing units according to the time dimension, and each unit independently calculates the local optimal probability value of each state at the current time step; adopts a pruning strategy to retain the state paths with the top k% probability values, and uses a bidirectional backtracking mechanism to perform bidirectional backtracking on the pruned state paths, and completes the integration of the optimal path at the intermediate intersection point, that is, obtains the optimal state path.

[0100] As an example, the optimization compensation unit 1002 specifically: calculates the fluctuation coefficient based on the raw material mixing uniformity and the preset mixing degree threshold, and calculates the entropy value of the state transition probability distribution at the current time step; calculates the k value based on the fluctuation coefficient and the entropy value, and uses a pruning strategy to retain the state paths with the first k% of the probability values.

[0101] This invention also provides an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program, when executed by the processor, implements the method as described in any of the foregoing embodiments.

[0102] This invention also provides a computer storage medium storing a computer program that can be executed by a processor to implement the methods described in any of the foregoing claims.

[0103] This invention also provides a computer program product comprising a computer program that can be executed by a processor to implement the method as described in any of the foregoing embodiments.

[0104] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0105] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered 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 the claims.

Claims

1. A method for quality control in the production of rubber cables, characterized in that: The method includes the following steps: acquiring raw material formulation data and raw material mixing data; the raw material formulation data includes the ratio data of rubber and its additives, and the raw material mixing data includes the mixing data of the batch of raw materials by the mixing equipment; predicting the raw material mixing uniformity based on the raw material mixing data, and generating interference sequence data based on the raw material mixing uniformity; processing the raw material formulation data and the interference sequence data using a multi-parameter collaborative optimization model to obtain target control parameters, including extrusion speed, extrusion temperature, and extrusion pressure; and controlling the rubber extrusion operation of the extrusion equipment according to the target control parameters.

2. The method for quality control in the production of rubber cables according to claim 1, characterized in that: The method for generating interference sequence data based on the uniformity of raw material mixing includes: determining whether the uniformity of raw material mixing is higher than a preset threshold; if so, then not generating interference sequence data based on the uniformity of raw material mixing; if not, then generating interference sequence data based on the uniformity of raw material mixing; wherein the preset threshold is derived from at least one of the following: rubber material characteristics, rubber cable performance requirements, and extrusion equipment performance.

3. The method for quality control in the production of rubber cables according to claim 1, characterized in that: Based on the raw material mixing data, the raw material mixing uniformity is predicted. Based on the raw material mixing uniformity, interference sequence data is generated, including: extracting multimodal features, including physical features, dynamic features, and temporal features, from the raw material mixing data; using an ensemble prediction model based on a stacked ensemble framework to predict the multimodal features to obtain the raw material mixing uniformity; wherein, the feature weights of each base model in the ensemble prediction model are dynamically adjusted based on the mixing duration; inputting the raw material mixing uniformity and the real-time observation sequence into a hidden Markov model, using the Viterbi algorithm to solve for the optimal state path, and calculating the state prediction distribution for the next n steps; wherein, the hidden Markov model defines multiple stages; using a stage-parameter mapping function, a dynamically adjusted sequence of each interference quantity changing over time is generated, i.e., the interference sequence data.

4. The method for quality control in the production of rubber cables according to claim 3, characterized in that: The Viterbi algorithm is used to solve for the optimal state path, which includes: dividing the state transition grid of the Hidden Markov Model into m parallel processing units according to the time dimension, with each unit independently calculating the local optimal probability value of each state at the current time step; using a pruning strategy to retain the top k% of the state paths with the highest probability values; and using a bidirectional backtracking mechanism to perform bidirectional backtracking on the pruned state paths, completing the integration of the optimal paths at the intermediate intersection point, thus obtaining the optimal state path.

5. The method for quality control in the production of rubber cables according to claim 4, characterized in that: The process of retaining the top k% of state paths with the highest probability values ​​using a pruning strategy includes: calculating a fluctuation coefficient based on the raw material mixing uniformity and a preset mixing degree threshold, and calculating the entropy value of the state transition probability distribution at the current time step; calculating a k value based on the fluctuation coefficient and the entropy value, and retaining the top k% of state paths with the highest probability values ​​using a pruning strategy.

6. A quality control system for rubber cable production, characterized in that, The system includes an acquisition unit, an optimization and compensation unit, and a control unit. The acquisition unit is used to acquire raw material formula data and raw material mixing data. The raw material formula data includes the proportion data of rubber and its additives, and the raw material mixing data includes the mixing data of the mixing equipment for this batch of raw materials. The optimization and compensation unit is used to predict the raw material mixing uniformity based on the raw material mixing data and generate interference sequence data based on the raw material mixing uniformity. The raw material formulation data and the interference sequence data are processed using a multi-parameter collaborative optimization model to obtain target control parameters, including extrusion speed, extrusion temperature, and extrusion pressure; the control unit is used to control the rubber extrusion operation of the extrusion equipment according to the target control parameters.

7. A rubber cable production quality control system according to claim 6, characterized in that: The optimization compensation unit specifically determines whether the uniformity of the raw material mixing is higher than a preset threshold. If so, it does not generate interference sequence data based on the uniformity of the raw material mixing; otherwise, it generates interference sequence data based on the uniformity of the raw material mixing. The preset threshold is derived based on at least one of the following: rubber material characteristics, rubber cable performance requirements, and extrusion equipment performance.

8. An electronic device, characterized in that: The electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program, when executed by the processor, implements the method as described in any one of claims 1-5.

9. A computer storage medium, characterized in that: The computer storage medium stores a computer program that can be executed by a processor to implement the method as described in any one of claims 1-5.

10. A computer program product, characterized in that: The computer program product includes a computer program that can be executed by a processor to implement the method as described in any one of claims 1-5.

Images