An energy-saving optimization method for cloth setting machine based on artificial intelligence

Abstract

The present invention relates to an energy-saving optimization method for a cloth setting machine based on artificial intelligence, comprising the following steps: S1: collecting process operation data in the cloth setting process; S2: preprocessing the collected data to eliminate noise and anomalies; S3: dividing the setting machine control system into a heating control subsystem, a dehumidification control subsystem and a circulating wind control subsystem according to different functions; S4: according to the predicted energy consumption demand, using a multi-objective optimization algorithm, optimizing the two goals of setting quality and energy consumption at the same time, and outputting the best temperature, humidity, and air flow speed combination curve scheme; S5: introducing deep reinforcement learning to train an intelligent controller, directly feedback the setting machine working conditions through real-time data, and dynamically adjusting the operation of each subsystem based on the best temperature, humidity, and air flow speed combination curve scheme. The present invention significantly improves the cloth setting quality and energy utilization efficiency.

Description

Technical Field

[0001] The present invention relates to the field of energy consumption optimization, and in particular to an energy-saving optimization method for a cloth setting machine based on artificial intelligence. Background Art

[0002] Fabric setting machines are key equipment in the textile industry, used to finish and process fabrics during the heating, stretching and setting process to ensure that they meet the final physical performance requirements, such as stable size, flat appearance and uniform density. However, the operation process of traditional fabric setting machines generally has problems such as high energy consumption, low efficiency and difficulty in accurately controlling process parameters. In particular, when there is a strong coupling relationship between core process parameters such as temperature, humidity, and air flow speed, and when the fabric material, fabric speed and working conditions change in real time, it is difficult to achieve a balance between high-quality production and energy-saving needs through conventional means. Summary of the invention

[0003] In order to solve the above problems, the purpose of the present invention is to provide an energy-saving optimization method for a cloth setting machine based on artificial intelligence, which significantly improves the cloth setting quality and energy utilization efficiency.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] An energy-saving optimization method for a cloth setting machine based on artificial intelligence comprises the following steps:

[0006] S1: Collect process operation data in the cloth shaping process, including temperature, humidity, pressure, cloth speed and tension data;

[0007] S2: pre-process the collected temperature, humidity, pressure, cloth speed and tension data to eliminate noise and anomalies;

[0008] S3: Divide the setting machine control system into a heating control subsystem, a dehumidification control subsystem and a circulating air control subsystem according to different functions;

[0009] S4: Based on the predicted energy consumption demand, a multi-objective optimization algorithm is used to simultaneously optimize the two objectives of shaping quality and energy consumption, and output the best temperature, humidity, and airflow velocity combination curve solution;

[0010] S5: Introduce deep reinforcement learning to train the intelligent controller, directly feedback the working conditions of the molding machine through real-time data, and dynamically adjust the operation of each subsystem based on the optimal temperature, humidity, and air flow speed combination curve solution.

[0011] Furthermore, S1 is specifically: introducing high-precision sensors in each operating link of the cloth setting machine to collect process operation data in the cloth setting process, including: temperature sensors, installed in each heating zone inside the setting machine, real-time monitoring of the temperature of each zone, and detection of heating uniformity; humidity sensors, installed at the dehumidification air flow outlet, real-time detection of moisture emissions; capturing the residual moisture data of the cloth itself to determine the actual working conditions of drying; pressure sensors, used to monitor the pressure trends of the steam system and the circulating air system; tension sensors, used to monitor the tension changes of the cloth during the setting process; cloth speed sensors, detecting the cloth speed at the inlet and outlet ends of the cloth.

[0012] Further, S2 is specifically:

[0013] The collected process data include: temperature T(t), humidity H(t), pressure P(t), cloth speed V b (t), tension F b (t); Set the data collection frequency of the production process of the setting machine to f s ;

[0014] The original collected data is a time series matrix X, where the i-th data point X i =[T(t),H(t),P(t),V b (t),F b (t)];

[0015] Signal smoothing and filtering methods are used to denoise the original collected data. Kalman filtering is used to denoise the pressure, cloth speed and tension data, and wavelet denoising is used to denoise the temperature and humidity data.

[0016] Based on the deviation detection of normal distribution, the Z score is defined as:

[0017] ;

[0018] Where μ is the sample mean; σ is the sample standard deviation; when |Z i ∣>Z threshold , the point is considered as an outlier and removed. threshold is the preset threshold;

[0019] Finally, the data is normalized.

[0020] Furthermore, Kalman filtering is used to denoise the pressure, cloth speed, and tension data, as follows:

[0021] Define state transition models for pressure, fabric speed or tension X k :

[0022] ;

[0023] in, x k is the current state; is the state change rate;

[0024] ;

[0025] ;

[0026] Where A is the state transfer matrix; is the sampling interval; B is the control matrix; u k is the control input; w k is the process noise;

[0027] The actual data collected is obtained by measuring the system status:

[0028] ;

[0029] Among them, z k is the observed data; H is the measurement matrix; v k To measure noise;

[0030] The prediction process includes state prediction and error covariance prediction:

[0031] ;

[0032] ;

[0033] in, is the estimated value of the predicted state at time k; is the prediction error covariance matrix; Q is the process noise covariance matrix;

[0034] Update process:

[0035] Kalman Gain K k calculate:

[0036] ;

[0037] Where R is the measurement noise covariance matrix;

[0038] Status Update:

[0039] ;

[0040] Error covariance update:

[0041] ;

[0042] Where I is the identity matrix;

[0043] The prediction-update process is executed cyclically to output the denoised pressure, fabric speed and tension;

[0044] Furthermore, wavelet denoising is used for the temperature and humidity data, as follows:

[0045] Perform multi-scale decomposition on the input signal x(t):

[0046] ;

[0047] Among them, An(t) is the nth order approximate component; D j (t) is the j-th order detail component;

[0048] Wavelet decomposition is achieved through recursive filtering:

[0049] ;

[0050] in, , are low-pass and high-pass filters of wavelet basis functions; is the filter length;

[0051] For high frequency detail components D j (t) Perform threshold processing to remove noise:

[0052] ;

[0053] in, T j is the threshold size;

[0054] Each detail component after threshold processing Approximate low frequency component A n (t) Reconstructed into denoised signal:

[0055] .

[0056] Furthermore, the control system of the setting machine is divided into a heating control subsystem, a dehumidification control subsystem and a circulating air control subsystem according to different functions.

[0057] The heating control subsystem divides the heating area into a preheating area, a main heating area and a stabilization area to achieve different temperature curve targets respectively; dynamically adjusts the heating power of different areas based on the fabric speed, material and target temperature; and introduces a waste heat recovery device to improve energy efficiency by using the waste gas and dehumidification heat generated during the heating process;

[0058] The dehumidification control subsystem monitors the surface humidity of the cloth in real time and dynamically adjusts the dehumidification air volume;

[0059] The circulating air control subsystem adjusts the circulating air speed in different areas in real time, giving priority to the wind speed requirement of the main heating area, and controls the overall air volume to balance the fabric surface temperature based on the fabric speed and target temperature gradient. The circulating air speed is set for different processing areas to reduce the air volume in non-critical areas.

[0060] Furthermore, the heating control subsystem is controlled as follows:

[0061] Set the preheating zone temperature target T pre ; Main heating zone temperature target Tmain; Stable zone temperature target T stable ;

[0062] Automatically adjust the heating power of each area according to the process parameters so that the fabric reaches the target temperature in each area. The power dynamic adjustment formula is:

[0063] ;

[0064] in, For the Heating power required for the area; m b is the mass flow rate of the cloth passing through the heating area per unit time; c p Specific heat capacity of cloth; For the Zone target temperature; For cloth to enter Temperature in the area; For heating efficiency; The heat required for water evaporation;

[0065] Based on the above formula, the heat power supply of each heating area is dynamically adjusted;

[0066] Introduce a waste heat recovery system to recover sensible heat from exhaust gas in the heating area through a heat exchanger, which is used to preheat the heating air or heat the dry air in the wet exhaust air. The steam exhaust is condensed and refluxed and reused for heating. Utilize waste heat in a graded manner according to the temperature gradient to improve energy utilization.

[0067] Assume the exhaust gas temperature is T waste The recovered heat energy is used to heat the fresh air T fresh Temperature rise, then the heat recovery calculation formula is:

[0068] ;

[0069] Among them, Q recovered is the recovered heat; η recovery is the waste heat recovery efficiency; m waste is the exhaust gas mass flow rate; is the specific heat capacity of air; T ambientis the ambient temperature.

[0070] Further, the circulating air control subsystem is as follows:

[0071] In the main heating area, wind speed is prioritized to enhance heat transfer, improve heating uniformity and efficiency; in the preheating area and stabilization area: maintain the required minimum wind speed to minimize wind energy consumption;

[0072] Wind speed requirement of main heating area:

[0073] ;

[0074] in, is the specific heat capacity of air, is the air density; The circulation wind speed of the main heating area; The temperature difference of air flow in the main heating area; is the required heat transfer rate;

[0075] According to different heating zones, formulate dynamic wind speed distribution ratio:

[0076] ;

[0077] in, No. Regional distribution of wind speed; is the total air volume of the circulating fan; It is the wind speed distribution coefficient, which is dynamically adjusted according to the real-time temperature and heating target;

[0078] Based on the regional temperature gradient, the circulating air volume of each area is dynamically adjusted, and temperature balance is achieved through PID control:

[0079] ;

[0080] in, Temperature adjustment amount; is the actual temperature value;

[0081] PID air volume adjustment formula:

[0082] ;

[0083] in, Air volume adjustment value, , and They are proportional gain, integral gain and differential gain respectively.

[0084] Furthermore, S4 is specifically:

[0085] definition Regional Optimization Decision Variable: Temperature Profile ; Humidity curve ; Circulation wind speed curve ;

[0086] For each zone, the preheating zone, main heating zone, and stabilization zone are optimized separately:

[0087] The final multi-objective optimization problem is:

[0088] ;

[0089] ;

[0090] ;

[0091] Among them, f quality is the final quality objective function, w temp ,w humidity To adjust the temperature and humidity weight parameters; f 1 is temperature uniformity; f2 For humidity control; f energy Energy consumption objective function; f total Overall objective function;

[0092] The multi-objective optimization algorithm NSGA-II is used to solve the above problem and output the Pareto optimal solution set.

[0093] Furthermore, S5 is specifically:

[0094] In reinforcement learning, the intelligent controller is modeled as an agent whose interaction object is the environment of the stereotyped machine. The problem is modeled by the Markov decision process, which is defined as follows:

[0095] State space S: includes the real-time collected data of the molding machine working condition, which is used to describe the current state of the system:

[0096] S={T actual ,H actual ,V actual ,T target ,H target ,V target ,ΔT,ΔH,ΔV,E consumed}

[0097] Among them, T actual ,H actual ,V actual are the actual temperature, humidity and wind speed of each partition at present; T target ,H target ,V targetare the target temperature, humidity, and wind speed respectively; the difference between the actual and target values ​​of ΔT, ΔH, and ΔV; and E consumed is the total energy consumption of the current system.

[0098] Action space A: includes adjustable control instructions for each subsystem:

[0099] A={P heat ,V exhaust ,V circulation};

[0100] Among them, P heat is the heating power; V exhaust V is the dehumidification wind speed; circulation is the circulating wind speed.

[0101] The reward function R takes into account the finalization quality reward R quality and energy consumption reward R energy :

[0102] ;

[0103] Among them, α is the balance weight between energy consumption and quality;

[0104] Based on the PPO deep reinforcement learning intelligent controller, the policy gradient method is used to directly output the control action a through a parameterized policy πθ(a|s), and the policy is updated under the specified constraints to improve performance;

[0105] Through real-time prediction and dynamic optimization:

[0106] Heating control subsystem: By adjusting the heating power P heat (t), to achieve operating temperature T actual With target T target Quick matching, while reducing unnecessary heating energy consumption;

[0107]

[0108] Dehumidification control subsystem: dynamically adjust the dehumidification wind speed V exhaust (t) To precisely control the humidity of the fabric surface;

[0109]

[0110] Circulation air control subsystem: real-time adjustment of partition circulation air speed V circulation (t) To ensure uniform heating and temperature and humidity consistency in each area:

[0111] .

[0112] The present invention has the following beneficial effects:

[0113] 1. The present invention solves the problems of high energy consumption, low efficiency and difficult quality control in the production process of the cloth setting machine, and significantly improves the cloth setting quality and energy utilization efficiency;

[0114] 2. In the process data processing, the present invention adopts a targeted filtering method for different data characteristics and noise types, which can more effectively eliminate noise, improve signal quality and processing efficiency, select a suitable denoising algorithm according to data characteristics, and retain the useful information of the original signal to the maximum extent, while avoiding information loss or signal distortion caused by inappropriate filtering methods;

[0115] 3. The present invention realizes efficient, dynamic and coordinated regional control and energy consumption optimization by combining the heating control subsystem, the dehumidification control subsystem and the circulating air control subsystem, which can accurately meet the needs of cloth shaping, comprehensively reduce energy consumption, and ensure the uniformity and stability of cloth quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0116] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0117] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0118] refer to Figure 1 In this embodiment, a method for energy-saving optimization of a cloth setting machine based on artificial intelligence is provided, comprising the following steps:

[0119] S1: Collect process operation data in the cloth shaping process, including temperature, humidity, pressure, cloth speed and tension data;

[0120] S2: pre-process the collected temperature, humidity, pressure, cloth speed and tension data to eliminate noise and anomalies;

[0121] S3: Divide the setting machine control system into a heating control subsystem, a dehumidification control subsystem and a circulating air control subsystem according to different functions;

[0122] S4: Based on the predicted energy consumption demand, a multi-objective optimization algorithm is used to simultaneously optimize the two objectives of shaping quality and energy consumption, and output the best temperature, humidity, and airflow velocity combination curve solution;

[0123] S5: Introduce deep reinforcement learning to train the intelligent controller, directly feedback the working conditions of the molding machine through real-time data, and dynamically adjust the operation of each subsystem based on the optimal temperature, humidity, and air flow speed combination curve solution.

[0124] In this embodiment, S1 specifically refers to: introducing high-precision sensors in each operating link of the cloth setting machine to collect process operation data in the cloth setting process, including: temperature sensors, installed in each heating zone inside the setting machine, real-time monitoring of the temperature of each zone, and detection of heating uniformity; humidity sensors, installed at the dehumidification air flow outlet, real-time detection of moisture emissions; capturing the residual moisture data of the cloth itself to determine the actual working conditions of drying; pressure sensors, used to monitor the pressure trends of the steam system and the circulating air system; tension sensors, used to monitor the tension changes of the cloth during the setting process; cloth speed sensors, detecting the cloth speed at the inlet and outlet ends of the cloth.

[0125] In this embodiment, S2 is specifically:

[0126] The collected process data include: temperature T(t), humidity H(t), pressure P(t), cloth speed V b (t), tension F b (t); Set the data collection frequency of the production process of the setting machine to f s ;

[0127] The original collected data is a time series matrix X, where the i-th data point X i =[T(t),H(t),P(t),V b (t),F b (t)];

[0128] Signal smoothing and filtering methods are used to denoise the original collected data. Kalman filtering is used to denoise the pressure, cloth speed and tension data, and wavelet denoising is used to denoise the temperature and humidity data.

[0129] Based on the deviation detection of normal distribution, the Z score is defined as:

[0130] ;

[0131] Where μ is the sample mean; σ is the sample standard deviation; when |Z i ∣>Z threshold , the point is considered as an outlier and removed. threshold is the preset threshold;

[0132] Finally, the data is normalized.

[0133] In this embodiment, Kalman filtering is used to denoise the pressure, cloth speed and tension data, as follows:

[0134] Define state transition models for pressure, fabric speed or tension X k :

[0135] ;

[0136] in, x k is the current state; State change rate; (such as pressure change rate, cloth speed change rate, tension change rate)

[0137] ;

[0138] ;

[0139] Where A is the state transfer matrix; is the sampling interval; B is the control matrix; u k is the control input (zero in the calculation of fabric speed and tension because the signal disturbance model ignores the external control effect); k is the process noise;

[0140] The actual data collected is obtained by measuring the system status:

[0141] ;

[0142] Among them, z k is the observed data (with noise); H is the measurement matrix; v k To measure noise;

[0143] The prediction process includes state prediction and error covariance prediction:

[0144] ;

[0145] ;

[0146] in, is the estimated value of the predicted state at time k; is the prediction error covariance matrix; Q is the process noise covariance matrix;

[0147] Update process:

[0148] Kalman Gain K k calculate:

[0149] ;

[0150] Where R is the measurement noise covariance matrix;

[0151] Status Update:

[0152] ;

[0153] Error covariance update:

[0154] ;

[0155] Where I is the identity matrix;

[0156] The prediction-update process is executed cyclically to output the denoised pressure, fabric speed and tension;

[0157] In this embodiment, wavelet denoising is used for temperature and humidity data, as follows:

[0158] Perform multi-scale decomposition on the input signal x(t) (temperature or humidity):

[0159] ;

[0160] Among them, A n (t) is the nth order approximate component; D j (t) is the j-th order detail component;

[0161] Wavelet decomposition is achieved through recursive filtering:

[0162] ;

[0163] in, , are low-pass and high-pass filters of wavelet basis functions; is the filter length;

[0164] For high frequency detail components D j (t) Perform threshold processing to remove noise:

[0165] ;

[0166] in, T j is the threshold size;

[0167] Each detail component after threshold processing The low frequency approximation component A n (t) is reconstructed into the denoised signal X denoised (t):

[0168] .

[0169] In this embodiment, the setting machine control system is divided into a heating control subsystem, a dehumidification control subsystem and a circulating air control subsystem according to different functions.

[0170] The heating control subsystem divides the heating area into a preheating area, a main heating area and a stabilization area to achieve different temperature curve targets respectively; dynamically adjusts the heating power of different areas based on the fabric speed, material and target temperature; and introduces a waste heat recovery device to improve energy efficiency by using the waste gas and dehumidification heat generated during the heating process;

[0171] The dehumidification control subsystem monitors the surface humidity of the cloth in real time and dynamically adjusts the dehumidification air volume;

[0172] The circulating air control subsystem adjusts the circulating air speed in different areas in real time, giving priority to the wind speed requirement of the main heating area, and controls the overall air volume to balance the fabric surface temperature based on the fabric speed and target temperature gradient. The circulating air speed is set for different processing areas to reduce the air volume in non-critical areas.

[0173] In this embodiment, the heating control subsystem controls are specifically as follows:

[0174] Set the preheating zone temperature target T pre (e.g. 80°C - 120°C, depending on the moisture content and characteristics of the fabric); the temperature target of the main heating zone Tmain, which is dynamically adjusted according to processing requirements (e.g. 180°C - 220°C, which is set according to the fabric shaping requirements); the temperature target of the stabilization zone T stable , slightly lower than the main heating zone temperature to maintain stability (e.g. 170°C - 200°C);

[0175] Automatically adjust the heating power of each area according to the process parameters so that the fabric reaches the target temperature in each area. The power dynamic adjustment formula is:

[0176] ;

[0177] in, For the Heating power required for the area; m b is the mass flow rate of the cloth passing through the heating area per unit time; c p Specific heat capacity of cloth; For the Zone target temperature; For cloth to enter Temperature in the area; For heating efficiency; The heat required for water evaporation;

[0178] Based on the above formula, the heat power supply of each heating area is dynamically adjusted;

[0179] Introduce a waste heat recovery system to recover sensible heat from exhaust gas in the heating area through a heat exchanger, which is used to preheat the heating air or heat the dry air in the wet exhaust air. The steam exhaust is condensed and refluxed and reused for heating. Utilize waste heat in a graded manner according to the temperature gradient to improve energy utilization.

[0180] Assume the exhaust gas temperature is T waste The recovered heat energy is used to heat the fresh air T fresh Temperature rise, then the heat recovery calculation formula is:

[0181] ;

[0182] Among them, Q recovered is the recovered heat; η recovery is the waste heat recovery efficiency; m waste is the exhaust gas mass flow rate; is the specific heat capacity of air; T ambient is the ambient temperature.

[0183] In this embodiment, the circulating air control subsystem is as follows:

[0184] In the main heating area, wind speed is prioritized to enhance heat transfer, improve heating uniformity and efficiency; in the preheating area and stabilization area: maintain the required minimum wind speed to minimize wind energy consumption;

[0185] Wind speed requirement of main heating area:

[0186] ;

[0187] in, is the specific heat capacity of air, is the air density; The circulation wind speed of the main heating area; The temperature difference of air flow in the main heating area; is the required heat transfer rate;

[0188] According to different heating zones, formulate dynamic wind speed distribution ratio:

[0189] ;

[0190] in, No. Regional distribution of wind speed; is the total air volume of the circulating fan; It is the wind speed distribution coefficient, which is dynamically adjusted according to the real-time temperature and heating target;

[0191] Based on the regional temperature gradient, the circulating air volume of each area is dynamically adjusted, and temperature balance is achieved through PID control:

[0192] ;

[0193] in, Temperature adjustment amount; is the actual temperature value;

[0194] PID air volume adjustment formula:

[0195] ;

[0196] in, Air volume adjustment value, , and They are proportional gain, integral gain and differential gain respectively.

[0197] In this embodiment, S4 is specifically:

[0198] definition Regional Optimization Decision Variable: Temperature Profile ; Humidity curve ; Circulation wind speed curve ;

[0199] For each partition, the preheating zone, main heating zone, and stabilization zone are optimized separately;

[0200] The final multi-objective optimization problem is:

[0201] ;

[0202] ;

[0203] ;

[0204] Among them, f quality is the final quality objective function, w temp ,w humidity To adjust the temperature and humidity weight parameters; f 1 is temperature uniformity; f2 For humidity control; f energy Energy consumption objective function; f total Overall objective function;

[0205] The multi-objective optimization algorithm NSGA-II is used to solve the above problem and output the Pareto optimal solution set.

[0206] In this embodiment, S5 is specifically:

[0207] In reinforcement learning, the intelligent controller is modeled as an agent whose interaction object is the environment of the stereotyped machine. The problem is modeled by the Markov decision process, which is defined as follows:

[0208] State space S: includes the real-time collected data of the molding machine working condition, which is used to describe the current state of the system:

[0209] S={T actual ,H actual ,V actual ,T target ,H target ,V target ,ΔT,ΔH,ΔV,E consumed}

[0210] Among them, T actual ,H actual ,V actual are the actual temperature, humidity and wind speed of each partition at present; T target ,H target ,V target are the target temperature, humidity, and wind speed respectively; the difference between the actual and target values ​​of ΔT, ΔH, and ΔV; and E consumed is the total energy consumption of the current system.

[0211] Action space A: includes adjustable control instructions for each subsystem:

[0212] A={P heat ,V exhaust ,V circulation};

[0213] Among them, P heat is the heating power; V exhaust V is the dehumidification wind speed; circulation is the circulating wind speed.

[0214] The reward function R takes into account the finalization quality reward R quality and energy consumption reward R energy :

[0215] ;

[0216] Among them, α is the balance weight between energy consumption and quality;

[0217] Based on the PPO deep reinforcement learning intelligent controller, the policy gradient method is used to directly output the control action a through a parameterized policy πθ(a|s), and the policy is updated under the specified constraints to improve the performance;

[0218] Through real-time prediction and dynamic optimization:

[0219] Heating control subsystem: By adjusting the heating power Pheat (t), to achieve operating temperature T actual With target T target Quick matching, while reducing unnecessary heating energy consumption;

[0220]

[0221] Dehumidification control subsystem: dynamically adjust the dehumidification wind speed V exhaust (t) To precisely control the humidity of the fabric surface;

[0222]

[0223] Circulation air control subsystem: real-time adjustment of partition circulation air speed V circulation (t) To ensure uniform heating and temperature and humidity consistency in each area:

[0224] .

[0225] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0226] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0227] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0228] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0229] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. An energy-saving optimization method for a cloth setting machine based on artificial intelligence, characterized in that: The following steps are involved: S1: Collect process operation data in the cloth shaping process, including temperature, humidity, pressure, cloth speed and tension data; S2: pre-process the collected temperature, humidity, pressure, cloth speed and tension data to eliminate noise and anomalies; S3: Divide the setting machine control system into a heating control subsystem, a dehumidification control subsystem and a circulating air control subsystem according to different functions; S4: Based on the predicted energy consumption demand, a multi-objective optimization algorithm is used to simultaneously optimize the two objectives of shaping quality and energy consumption, and output the best temperature, humidity, and airflow velocity combination curve solution; S5: Introduce deep reinforcement learning to train the intelligent controller, directly feedback the working conditions of the molding machine through real-time data, and dynamically adjust the operation of each subsystem based on the optimal temperature, humidity, and air flow speed combination curve scheme; The control system of the setting machine is divided into a heating control subsystem, a dehumidification control subsystem and a circulating air control subsystem according to different functions. The heating control subsystem divides the heating area into a preheating area, a main heating area and a stabilization area to achieve different temperature curve targets respectively; dynamically adjusts the heating power of different areas based on the fabric speed, material and target temperature; and introduces a waste heat recovery device to improve energy efficiency by using the waste gas and dehumidification heat generated during the heating process; The dehumidification control subsystem monitors the surface humidity of the cloth in real time and dynamically adjusts the dehumidification air volume; The circulating wind control subsystem adjusts the circulating wind speed of different areas in real time, giving priority to the wind speed requirement of the main heating area, and controls the overall air volume to balance the fabric surface temperature based on the fabric speed and target temperature gradient; sets the circulating wind speed for different processing areas and reduces the air volume in non-critical areas; The heating control subsystem controls are specifically as follows: Set the preheating zone temperature target T pre ; Main heating zone temperature target Tmain; Stable zone temperature target T stable ; Automatically adjust the heating power of each area according to the process parameters so that the fabric reaches the target temperature in each area. The power dynamic adjustment formula is: ； in, For the Heating power required for the area; m b is the mass flow rate of the cloth passing through the heating area per unit time; c p Specific heat capacity of cloth; For the Zone target temperature; For cloth to enter Temperature in the area; For heating efficiency; is the heat required for water evaporation; based on the above formula, the heat power supply of each heating area is dynamically adjusted.

2. The energy-saving optimization method for a cloth setting machine based on artificial intelligence according to claim 1 is characterized in that: The S1 is specifically: introducing high-precision sensors in each operating link of the cloth setting machine to collect process operation data in the cloth setting process, including: temperature sensors, installed in each heating zone inside the setting machine, real-time monitoring of the temperature of each zone, and detection of heating uniformity; humidity sensors, installed at the dehumidification air flow outlet, real-time detection of moisture emission; capturing the residual moisture data of the cloth itself to determine the actual working condition of drying; pressure sensors, used to monitor the pressure trend of the steam system and the circulating air system; tension sensors, used to monitor the tension changes of the cloth during the setting process; cloth speed sensors, detecting the cloth speed at the inlet and outlet ends of the cloth.

3. The energy-saving optimization method for a cloth setting machine based on artificial intelligence according to claim 1 is characterized in that: The S2 is specifically: The collected process data include: temperature T(t), humidity H(t), pressure P(t), cloth speed V b (t), tension F b (t); Set the data collection frequency of the production process of the setting machine to f s ; The original collected data is a time series matrix X, where the i-th data point X i =[T(t),H(t),P(t),V b (t),F b (t)]; Signal smoothing and filtering methods are used to denoise the original collected data. Kalman filtering is used to denoise the pressure, cloth speed and tension data, and wavelet denoising is used to denoise the temperature and humidity data. Based on the deviation detection of normal distribution, the Z score is defined as: ； Where μ is the sample mean; σ is the sample standard deviation; when |Z i ∣>Z threshold , the point is considered as an outlier and removed. threshold is the preset threshold; Finally, the data is normalized.

4. The energy-saving optimization method for a cloth setting machine based on artificial intelligence according to claim 3 is characterized in that: The pressure, cloth speed and tension data are denoised using Kalman filtering, as follows: Define state transition models for pressure, fabric speed or tension X k : ； in, x k is the current state; is the state change rate; ； ； Where A is the state transfer matrix; is the sampling interval; B is the control matrix; u k is the control input; w k is the process noise; The actual data collected is obtained by measuring the system status: ； Among them, z k is the observed data; H is the measurement matrix; v k To measure noise; The prediction process includes state prediction and error covariance prediction: ； ； in, is the estimated value of the predicted state at time k; is the prediction error covariance matrix; Q is the process noise covariance matrix; Update process: Kalman Gain K k calculate: ； Where R is the measurement noise covariance matrix; Status Update: ； Error covariance update: ； Where I is the identity matrix; The prediction-update process is executed cyclically to output the denoised pressure, fabric speed and tension.

5. The energy-saving optimization method for cloth setting machine based on artificial intelligence according to claim 3 is characterized in that: The temperature and humidity data are denoised using wavelet, as follows: Perform multi-scale decomposition on the input signal x(t): ; Among them, An(t) is the nth order approximate component; D j (t) is the j-th order detail component; Wavelet decomposition is achieved through recursive filtering: ； in, , are low-pass and high-pass filters of wavelet basis functions; is the filter length; For high frequency detail components D j (t) Perform threshold processing to remove noise: ； in, T j is the threshold size; Each detail component after threshold processing Approximate low frequency component A n (t) is reconstructed into the denoised signal X denoised (t): 。 6. The energy-saving optimization method for cloth setting machine based on artificial intelligence according to claim 1 is characterized in that: Introduce a waste heat recovery system to recover sensible heat from exhaust gas in the heating area through a heat exchanger, which is used to preheat the heating air or heat the dry air in the wet exhaust air. The steam exhaust is condensed and refluxed and reused for heating. Utilize waste heat in a graded manner according to the temperature gradient to improve energy utilization. Assume the exhaust gas temperature is T waste The recovered heat energy is used to heat the fresh air T fresh Temperature rise, then the heat recovery calculation formula is: ； Among them, Q recovered is the recovered heat; η recovery is the waste heat recovery efficiency; m waste is the exhaust gas mass flow rate; is the specific heat capacity of air; T ambient is the ambient temperature.

7. The energy-saving optimization method for cloth setting machine based on artificial intelligence according to claim 1 is characterized in that: The circulating air control subsystem is specifically as follows: In the main heating area, wind speed is prioritized to enhance heat transfer, improve heating uniformity and efficiency; in the preheating area and stabilization area: maintain the required minimum wind speed to minimize wind energy consumption; Wind speed requirement of main heating area: ； in, is the specific heat capacity of air, is the air density; The circulation wind speed of the main heating area; The temperature difference of airflow in the main heating area; is the required heat transfer rate; According to different heating zones, formulate dynamic wind speed distribution ratio: ； in, No. Regional distribution of wind speed; is the total air volume of the circulating fan; It is the wind speed distribution coefficient, which is dynamically adjusted according to the real-time temperature and heating target; Based on the regional temperature gradient, the circulating air volume of each area is dynamically adjusted, and temperature balance is achieved through PID control: ； in, Temperature adjustment amount; is the actual temperature value; PID air volume adjustment formula: ； in, Air volume adjustment value, , and They are proportional gain, integral gain and differential gain respectively.

8. The energy-saving optimization method for a cloth setting machine based on artificial intelligence according to claim 1 is characterized in that: The S4 is specifically: definition Regional Optimization Decision Variable: Temperature Profile ; Humidity curve ; Circulation wind speed curve ; For each zone, the preheating zone, main heating zone, and stabilization zone are optimized separately: The final multi-objective optimization problem is: ； ； ； Among them, f quality is the final quality objective function, w temp ,w humidity To adjust the temperature and humidity weight parameters; f 1 is temperature uniformity; f2 For humidity control; f energy Energy consumption objective function; f total Overall objective function; The multi-objective optimization algorithm NSGA-II is used to solve the above problem and output the Pareto optimal solution set.

9. The energy-saving optimization method for a cloth setting machine based on artificial intelligence according to claim 1 is characterized in that: The S5 is specifically: In reinforcement learning, the intelligent controller is modeled as an agent whose interaction object is the environment of the stereotyped machine. The problem is modeled by the Markov decision process, which is defined as follows: State space S: includes the real-time collected data of the molding machine working condition, which is used to describe the current state of the system: S={T actual ,H actual ,V actual ,T target ,H target ,V target ,ΔT,ΔH,ΔV,E consumed }; Among them, T actual ,H actual ,V actual are the actual temperature, humidity and wind speed of each partition at present; T target ,H target ,V target are the target temperature, humidity, and wind speed respectively; the difference between the actual and target values ​​of ΔT, ΔH, and ΔV; and E consumed is the total energy consumption of the current system; Action space A: includes adjustable control instructions for each subsystem: A={P heat ,V exhaust ,V circulation }; Among them, P heat is the heating power; V exhaust V is the dehumidification wind speed; circulation is the circulation wind speed; The reward function R takes into account the finalization quality reward R quality and energy consumption reward R energy : ； Among them, α is the balance weight between energy consumption and quality; Based on the PPO deep reinforcement learning intelligent controller, the policy gradient method is used to directly output the control action a through a parameterized policy πθ(a|s), and the policy is updated under the specified constraints to improve performance; Through real-time prediction and dynamic optimization: Heating control subsystem: By adjusting the heating power P heat (t), to achieve operating temperature T actual With target T target Quick matching, while reducing unnecessary heating energy consumption; ； Dehumidification control subsystem: dynamically adjust the dehumidification wind speed V exhaust (t) To precisely control the humidity of the fabric surface; ； Circulation air control subsystem: real-time adjustment of partition circulation air speed V circulation (t) To ensure uniform heating and temperature and humidity consistency in each area: 。

Images