Manufacturing system distributed energy consumption real-time control method considering uncertainty

Abstract

The invention discloses a manufacturing system distributed energy consumption real-time control method considering uncertainty, and belongs to the technical field of energy consumption control, the technical key points are that the method comprises the following steps: 1, simplifying a production system into a serial production line composed of M machines, the # imgabs0 # is the machine with the highest energy consumption and the slowest processing speed; 2, a real-time energy consumption control function considers the current energy consumption loss and the future potential energy consumption loss of the production system at the same time in a forward view window, and changes along with the change of the energy consumption state of the manufacturing system; 3, evaluating the corresponding energy of the Markov strategy by adopting a real-time energy consumption control function; step 4, a real-time energy consumption control function, wherein real-time distributed feedback control is integration of a real-time local control strategy and a real-time cooperative control strategy; and step 5, obtaining a control rule and implementing the scheme, and the method has the advantage that the energy efficiency of the manufacturing system is effectively improved by designing a real-time distributed feedback control strategy for a single machine.

Description

Technical Field

[0001] The present invention relates to the technical field of energy consumption control, and in particular to a real-time control method for distributed energy consumption of a manufacturing system taking uncertainty into consideration. Background Art

[0002] Traditionally, the main research focus of manufacturing systems has been on improving the efficiency, flexibility and responsiveness of production, while energy consumption at the system level has received little attention. Researchers have found that 85% of the energy consumption of a production plant is used for functions unrelated to part production, and a large amount of energy is wasted during machine blockages and downtime for waiting for materials. With the rise in global energy prices and the improvement of environmental awareness, it is particularly important to increase the competitiveness of production companies by reducing the energy consumption of manufacturing systems and improving energy efficiency in energy-intensive manufacturing processes. It is imperative to find ways to reduce energy waste in the processing of manufacturing systems.

[0003] Massive production process data of manufacturing systems are collected and stored in real time, but they are not currently used to improve energy efficiency, which is the key to improving the real-time control response capability of manufacturing systems, but is often overlooked. The massive time series data collected by sensors in energy management systems must be analyzed in real time and converted into usable knowledge for energy consumption control considering multidimensional uncertainties. In recent years, the use of sensor data for transient analysis of production systems has attracted increasing attention. Existing methods based on long-term steady-state assumptions are no longer applicable to real-time transient analysis and cannot fully utilize the huge advantages of distributed sensor networks to enable manufacturing systems to respond quickly to environmental changes. Many scholars have used readily available sensor information on production lines to study dynamic real-time production performance constraints. However, these studies focus on real-time system diagnosis without further predicting potential future system performance. Sensor data are mostly used for isolated local control, providing unprecedented opportunities for improving quality and productivity, but have not shown potential in explaining and evaluating overall system performance.

[0004] Manufacturing systems lack real-time energy consumption control methods for energy efficiency optimization. Although intelligent real-time control methods have a long history of research in the field of control theory, no relevant research has been carried out in the manufacturing field. Manufacturing system energy optimization lacks corresponding real-time intelligent control methods. Traditional manufacturing system control methods are mainly based on steady-state analysis and long-term performance measurement, lacking the prediction of the dynamic state of complex production systems at any time in the future, and the estimation of the long-term steady-state performance of the system cannot be used for real-time optimization control. In addition, energy control and energy management architecture methods have been developed and applied to improve the energy efficiency of the system, and a factory energy monitoring and management framework combining energy data exchange, online energy data analysis, performance measurement and energy usage display standards has been proposed.

[0005] Research on improving energy efficiency in manufacturing shop floor operations has received increasing attention. Previous work in this area has focused on isolated or mutually independent machines or processes. However, in complex hybrid manufacturing systems with dynamic continuity and multi-dimensional uncertain interruptions, the operating state of each machine is determined not only by itself but also by its upstream and downstream machines, showing highly complex dynamic characteristics. Therefore, considering the interaction of machines within the production system and multi-dimensional uncertain interruption events is essential for effective energy control.

[0006] Distributed methods are widely used in the design, evaluation and implementation of control structures for production systems. Distributed control refers to the division of a global control process into multiple decision entities that can support the decision process. The main reason for the industrial demand to move towards a modified hierarchical structure is that the industry needs to move from static optimality or near-optimality to dynamic optimization, agility and robustness.

[0007] In the production process of manufacturing systems, there are widespread multidimensional uncertainty interruption events, which may lead to production stoppage or equipment performance degradation, long idle time, serious energy waste and other problems. In order to ensure the normal production speed and quality of the manufacturing system, it is a great challenge to establish an energy consumption control strategy that can respond quickly to multidimensional uncertainty to reduce the total energy consumption of the manufacturing system and improve energy efficiency. Summary of the invention

[0008] In view of the deficiencies in the prior art, an embodiment of the present invention aims to provide a real-time control method for distributed energy consumption of a manufacturing system taking uncertainty into consideration, so as to solve the problems in the above-mentioned background technology.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] A real-time control method for distributed energy consumption of a manufacturing system considering uncertainty comprises the following steps:

[0011] Step 1: Simplify the production system into a serial production line consisting of M machines, where It is the machine with the highest energy consumption and the slowest processing speed;

[0012] Step 2: The real-time energy consumption control function considers both the current energy consumption loss and the future potential energy consumption loss of the production system in the “look-ahead window” and changes with the energy consumption state of the manufacturing system;

[0013] Step 3: Use the real-time energy consumption control function to evaluate the corresponding energy of the Markov strategy;

[0014] Step 4: Real-time energy consumption control function. Real-time distributed feedback control is the integration of real-time local control strategy and real-time collaborative control strategy. The local feedback strategy only considers the single machine S j Impact on energy consumption of manufacturing systems;

[0015] Step 5: Obtain control rules and implement the plan.

[0016] As a further solution of the present invention, in step 1

[0017] As a further solution of the present invention, the first machine S1 among the M machines will not wait for materials, and the last machine S M No blocking.

[0018] As a further solution of the present invention, each machine S in the M machines j Processing power K pr,j Equal to its rated power, the no-load power is K id,j , no energy is consumed when turned off.

[0019] As a further solution of the present invention, in step 3, W(t)=[W1(t),W2(t),...,W M (t)]′ is the multidimensional uncertain interruption event of the manufacturing system at time t, where W i (t) is used to describe the machine S i Whether a multi-dimensional uncertain disruption event is experienced at time t is directly related to E and is obtained from the sensor data in the energy manager.

[0020] As a further solution of the present invention, the step 4 also includes estimating the number of machines S in the manufacturing system. j The energy consumption of the two control options (on and off) is calculated, and the optimal energy consumption strategy is determined locally. j The real-time energy consumption control function is as follows:

[0021] C j (t) = C j,loc (t)+C j,col (t) (1)

[0022] Among them, C j,loc (t) represents machine S j Real-time local control of energy consumption, and C j,col (t) represents the machine S due to cooperative control j Energy consumption of real-time collaborative requests.

[0023] As a further solution of the present invention, the solution in step five is implemented based on the observed production data, and the mold change interval time and mold change time; production preparation interval time and production preparation time; parameter adjustment interval time and parameter adjustment time are all assumed to obey exponential distribution.

[0024] In summary, compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0025] 1. The proposed real-time distributed feedback energy consumption control strategy for manufacturing systems can achieve real-time control of the energy consumption of manufacturing systems and rapid response to multi-dimensional uncertain interruption events. It improves the accuracy and robustness of energy consumption control and improves the energy efficiency and profit of manufacturing systems;

[0026] 2. The proposed real-time distributed feedback energy consumption control strategy establishes a real-time energy consumption control function at the single machine level to determine whether the control input of each machine is on or off. This avoids the large amount of calculations required by a single central controller and greatly reduces the workload;

[0027] 3. The proposed energy consumption control strategy can quickly and accurately evaluate the real-time energy consumption and real-time energy loss of the manufacturing system, as well as predict the potential energy loss in the future, in order to find energy-saving opportunities.

[0028] 4. The proposed real-time distributed feedback energy consumption control strategy fully considers the interaction between machines within the manufacturing system and the multi-dimensional uncertain interruption events that occur during the production process, thus improving the effectiveness and robustness of the energy consumption control of the manufacturing system;

[0029] 5. The distributed feedback controller integrates the local control strategy and the collaborative control strategy. It only needs to evaluate the energy consumption of the two control options (on and off) of a single machine and determine the optimal strategy locally to achieve energy consumption control of the manufacturing system.

[0030] 6. The proposed real-time distributed feedback energy consumption control strategy can minimize the real-time control energy consumption and real-time energy consumption on the basis of ensuring normal production speed and quality without causing additional energy loss. It can also control the potential energy loss in the future and avoid frequent changes in the state of the machine.

[0031] In order to more clearly illustrate the structural features and effects of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a real-time energy consumption control process of an embodiment of the invention.

[0033] Figure 2 This is a real-time energy consumption control process of an embodiment of the invention.

[0034] Figure 3 It is a feedback control scheme for manufacturing system energy consumption in an embodiment of the invention. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0036] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0037] In one embodiment, a real-time control method for distributed energy consumption of a manufacturing system considering uncertainty is provided. Figure 1 to Figure 3 , including the following steps:

[0038] Step 1: Simplify the production system into a serial production line consisting of M machines, where It is the machine with the highest energy consumption and the slowest processing speed;

[0039] Step 2: The real-time energy consumption control function considers both the current energy consumption loss and the future potential energy consumption loss of the production system in the “look-ahead window” and changes with the energy consumption state of the manufacturing system;

[0040] Step 3: Use the real-time energy consumption control function to evaluate the corresponding energy of the Markov strategy;

[0041] Step 4: Real-time energy consumption control function. Real-time distributed feedback control is the integration of real-time local control strategy and real-time collaborative control strategy. The local feedback strategy only considers the single machine S j Impact on energy consumption of manufacturing systems;

[0042] Step 5: Obtain control rules and implement the plan.

[0043] For further information, see Figure 1 to Figure 3 , in step 1

[0044] For further information, see Figure 1 to Figure 3 , the first machine S1 among the M machines will not be waiting for material, and the last machine S M No blocking.

[0045] For further information, see Figure 1 to Figure 3 , each machine S in the M machines j Processing power K pr,j Equal to its rated power, the no-load power is K id,j , no energy is consumed when turned off.

[0046] For further information, see Figure 1 to Figure 3 In step 3, W(t)=[W1(t),W2(t),...,W M (t)]′ is the multidimensional uncertain interruption event of the manufacturing system at time t, where W i (t) is used to describe the machine S i Whether a multi-dimensional uncertain disruption event is experienced at time t is directly related to E and is obtained from the sensor data in the energy manager.

[0047] For further information, see Figure 1 to Figure 3 The step 4 also includes estimating the cost of each machine S in the manufacturing system. j The energy consumption of the two control options (on and off) is calculated, and the optimal energy consumption strategy is determined locally. j The real-time energy consumption control function is as follows:

[0048] C j (t) = C j,loc (t)+C j,col (t) (1)

[0049] Among them, C j,loc (t) represents machine S j Real-time local control of energy consumption, and C j,col (t) represents the machine S due to cooperative control j Energy consumption of real-time collaborative requests.

[0050] For further information, see Figure 1 to Figure 3 The implementation of the scheme in step five is based on the observed production data, and the mold change interval time and mold change time; production preparation interval time and production preparation time; parameter adjustment interval time and parameter adjustment time are all assumed to obey exponential distribution.

[0051] In this embodiment, the present invention studies a discrete event driven dynamic continuous three-stage bufferless production system with multi-dimensional uncertainty interruption and real-time energy consumption control strategy to achieve energy efficiency optimization. The production system is simplified into a serial production line consisting of M machines, where It is the machine with the highest energy consumption and the slowest processing speed, such as Figure 1 shown.

[0052] For the production system, we make the following assumptions:

[0053] 1) The energy consumption rate of each machine is 1 / T i , i=1,...,M, is proportional to the processing speed, where T i It's Machine S iProcessing cycle;

[0054] 2) Indicates the machine with the highest energy consumption and the slowest processing speed in the production line, that is,

[0055] 3) If the machine is running normally and its upstream machine is shut down, the machine is in the waiting state (starvation);

[0056] 4) If the machine is operating normally and the downstream machine is shut down, the machine is in a blocked state;

[0057] 5) The first machine S1 will never be waiting for material (starvation), and the last machine S M Never blocks;

[0058] 6) If the machine is running normally, neither waiting for material (starvation) nor blocked, it will run at processing power;

[0059] 7) Each machine S j Processing power K pr,j Equal to its rated power, the no-load power is K id,j , does not consume any energy when shut down;

[0060] 8) When a multi-dimensional uncertain interruption event occurs, the machine may be in a shut down or no-load state.

[0061] Based on the energy consumption mathematical model of the multi-dimensional uncertain manufacturing system, the present invention establishes a real-time distributed feedback control scheme for energy consumption based on the real-time local control energy consumption function and the real-time collaborative request energy consumption function, and designs a real-time distributed feedback control strategy for a single machine to effectively improve the energy efficiency of the manufacturing system. The control scheme integrates real-time feedback control and coordinated control, and integrates the current and predicted energy consumption status of the manufacturing system to achieve energy efficiency optimization.

[0062] The real-time energy consumption control process proposed by the present invention is as follows: Figure 2 As shown in the figure, the real-time energy consumption control function considers both the current energy consumption loss and the future potential energy consumption loss of the production system in the "look-ahead window", and changes with the change of the energy consumption state of the manufacturing system. For example, the energy consumption control function at time t1 and time t1+1 is different. When time t moves forward within a global time period [0, T], the control input will be continuously updated according to the real-time energy consumption control function at each local time instant.

[0063] The control framework proposed by the present invention is as follows Figure 3As shown, the control input U(t) = G(X(t), θ(t)) is responsible for switching each machine on or off. The feedback strategy G(*) depends on the current energy consumption state X(t) and the switch state θ(t) of the manufacturing system. It is a Markov strategy. The present invention uses a real-time energy consumption control function to evaluate the corresponding energy of the Markov strategy. W(t) = [W1(t), W2(t), ..., W M (t)]′ is the multidimensional uncertain interruption event of the manufacturing system at time t. i (t) is used to describe the machine S i Whether a multi-dimensional uncertain disruption event is experienced at time t is directly related to E and can be obtained from the sensor data in the energy manager.

[0064] The real-time distributed feedback control proposed in this invention is an integration of the real-time local control strategy and the real-time collaborative control strategy, in which the local feedback strategy only considers the single machine S j The impact on the energy consumption of the manufacturing system; while the collaborative control considers the machine S j Is there a risk of constraining the operation of its neighboring machines? In this way, we only need to evaluate the risk of each machine S in the manufacturing system. j The energy consumption of the two control options (on and off) is calculated, and the optimal energy consumption strategy is determined locally. j The real-time energy consumption control function is as follows:

[0065] C j (t) = C j,loc (t)+C j,col (t) (1)

[0066] Among them, C j,loc (t) represents machine S j Real-time local control of energy consumption, and C j,col (t) represents the machine S due to cooperative control j Energy consumption of real-time collaborative requests.

[0067] (1) Real-time local control of energy consumption C j,loc (t)

[0068] C j,loc (t) = C j,E (t)+C j,PL (t)+C j,PPL (t)+C j,sw (t) (2)

[0069] Among them, C j,loc (t) represents machine S j Real-time local control of energy consumption, C j,E (t) represents machine S jReal-time energy consumption, C j,PL (t) represents the j The real-time energy consumption loss caused by j,PPL (t) indicates that due to the machine S j The potential energy loss in the future caused by j,sw (t) represents machine S j Energy consumption of state transition at time t.

[0070] C j,E (t) = K j I t θ j (t) (3)

[0071] C j,E (t) is used to evaluate the machine S j The real-time energy consumption caused by the current consumption rate per unit time. j Indicates machine S j The power consumption rate, I t represents the length of unit time, θ j (t) indicates the on / off state of the machine. j Based on the energy loss rate per unit time The real-time energy loss caused by j,PL (t) is calculated as:

[0072]

[0073] Because real-time decisions will directly affect the future potential energy consumption loss of the manufacturing system, the energy consumption control function needs to consider the future potential energy consumption loss. However, there is currently no systematic method to calculate this future potential energy consumption loss, so the present invention selects a discount rate r based on the current production status and priority preference. j At time t, the machine has made every effort to avoid the future potential energy loss, so there is no need to consider the future potential energy loss when the machine is running. j,PPL (t) is expressed as

[0074]

[0075] Machine S j The switching process between on and off causes additional energy consumption, which is called state transition energy consumption. To evaluate the machine S j The energy consumption of state transition at time t is Indicates machine S j Stay in the current state θ j (t) is the length of time. Therefore, the energy consumption of state transition is Cj,sw (t) is:

[0076]

[0077] (2) Energy consumption of real-time collaborative requests C j,col (t)

[0078] Considering the interaction between each machine in the manufacturing system, the machine S j Energy consumption of real-time collaborative request C j,col (t) is

[0079]

[0080] 5. Control Laws

[0081] The following pseudo program is used to describe how to determine the control input u of each device j (t).

[0082] Procedure 1.1

[0083] 1. Given X(t) and W(t), initial setting and find

[0084] 2. Set machine number j = j-1, where j' is the machine number in the previous step and u has been derived j′ (t); find

[0085] 3. Repeat step 2 until the machine is derived u of all upstream machines j (t).

[0086] Procedure 1.2

[0087] 1. Given X(t) and W(t), initial setting and find

[0088] 2. Set machine number j = j + 1, where j ′ is the machine number in the previous step, and u has been exported j′ (t); find

[0089] 3. Repeat step 2 until the machine is derived u of all downstream machines j (t).

[0090] in Represents the machine with the highest energy consumption and the slowest running speed. After obtaining the sensor data of X(t) and W(t) from the manufacturing system energy management system, we can deduce u for all machines in the manufacturing system based on procedures 1.1 and 1.2. j (t).

[0091] 6. Implementation

[0092] The production parameters are shown in Table 1. Based on the observed production data, the mold change interval and mold change time; production preparation interval and production preparation time; parameter adjustment interval and parameter adjustment time are all assumed to obey exponential distribution.

[0093] Table 1 Production parameters of the twelve machines in the production line.

[0094]

[0095]

[0096] In this case study, the discount rate for future potential energy losses is chosen as r = 0.50, the collaboration rate is chosen as r2 = 0.50, and the length of the unit time is chosen as I t = 1min. The conversion energy consumption function of the machine ( The units are in minutes) are assumed to be a simple linear function:

[0097]

[0098]

[0099] 1.3 Simulation Results

[0100] The simulation duration is three weeks, i.e. 30240 minutes. The simulation results with a production volume of 300 t are shown in Table 2.

[0101] Table 2 Simulation results when the production volume is 300T.

[0102]

[0103] Table 3 Energy consumption performance improvement results of the production line control scheme when the production volume is 300T

[0104] Reduction of processing energy consumption (%) Reduction of no-load energy consumption (%) Reduction in total energy consumption in production (%) Increase in energy efficiency (%) average value 1.44 55.34 23.00 28.00

[0105] The working principle of the present invention is:

[0106] On the basis of the mathematical model of energy consumption of multi-dimensional uncertain manufacturing system, a real-time distributed feedback control method of energy consumption based on real-time local control energy consumption function and real-time collaborative request energy consumption function is established. A real-time distributed feedback control strategy is designed for a single machine to effectively improve the energy efficiency of the manufacturing system.

[0107] The real-time energy consumption control function considers both the current energy consumption loss and the future potential energy consumption loss of the production system in the “look-ahead window”, and changes with the energy consumption state of the manufacturing system. The control input will be continuously updated according to the real-time energy consumption control function at each local time instant.

[0108] Feedback control framework for energy consumption of manufacturing system. The control input U(t)=G(X(t),θ(t)) is responsible for switching each machine on or off. The feedback strategy G(*) depends on the current energy consumption state X(t) and the switch state θ(t) of the manufacturing system. It is a Markov strategy. The response of the Markov strategy is evaluated using the real-time energy consumption control function. W(t)=[W1(t),W2(t),...,W M (t)]′ is the multidimensional uncertain interruption event of the manufacturing system at time t, which can be obtained from the sensor data in the energy manager.

[0109] Real-time distributed feedback control is an integration of real-time local control strategy and real-time collaborative control strategy, in which the local feedback strategy only considers the single machine S j The impact on the energy consumption of the manufacturing system; while the collaborative control considers the machine S j Is there a risk of constraining the operation of machines in its neighborhood?

[0110] The real-time distributed feedback control method of energy consumption fully considers the interaction between machines within the manufacturing system and the multi-dimensional uncertain interruption events occurring in the production process, thus improving the effectiveness and robustness of the energy consumption control of the manufacturing system.

[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A real-time control method for distributed energy consumption of a manufacturing system considering uncertainty, characterized in that: The following steps are involved: Step 1: Simplify the production system into a serial production line consisting of M machines, where It is the machine with the highest energy consumption and the slowest processing speed; Step 2: The real-time energy consumption control function considers both the current energy consumption loss and the future potential energy consumption loss of the production system in the "look-ahead window" and changes with the energy consumption state of the manufacturing system; Step 3: Use the real-time energy consumption control function to evaluate the corresponding energy of the Markov strategy; Step 4: Real-time energy consumption control function. Real-time distributed feedback control is the integration of real-time local control strategy and real-time collaborative control strategy. The local feedback strategy only considers the single machine S j Impact on energy consumption of manufacturing systems; Step 5: Obtain control rules and implement the plan.

2. The real-time control method for distributed energy consumption of a manufacturing system considering uncertainty according to claim 1 is characterized in that: In step one 3. The real-time control method for distributed energy consumption of a manufacturing system considering uncertainty according to claim 2 is characterized in that: The first machine S1 among the M machines will not be waiting for materials, and the last machine S M No blocking.

4. The real-time control method for distributed energy consumption of a manufacturing system considering uncertainty according to claim 3 is characterized in that: Each machine S in the M machines j Processing power K pr,j Equal to its rated power, the no-load power is K id,j , no energy is consumed when turned off.

5. The real-time control method for distributed energy consumption of a manufacturing system considering uncertainty according to claim 1 is characterized in that: In step 3, W(t)=[W1(t),W2(t),...,W M (t)]′ is the multidimensional uncertain interruption event of the manufacturing system at time t, where W i (t) is used to describe the machine S i Whether a multi-dimensional uncertain disruption event is experienced at time t is directly related to E and is obtained from the sensor data in the energy manager.

6. The real-time control method for distributed energy consumption of a manufacturing system considering uncertainty according to claim 1 is characterized in that: The step 4 also includes estimating the S of each machine in the manufacturing system. j The energy consumption of the two control options (on and off) is calculated, and the optimal energy consumption strategy is determined locally. j The real-time energy consumption control function is as follows: C j (t)=C j,loc (t)+C j,col (t) (1) Among them, C j,loc (t) represents machine S j Real-time local control of energy consumption, and C j,col (t) represents the machine S due to cooperative control j Energy consumption of real-time collaborative requests.

7. The real-time control method for distributed energy consumption of a manufacturing system considering uncertainty according to claim 1 is characterized in that: The implementation of the scheme in step five is based on the observed production data, and the mold change interval time and mold change time; production preparation interval time and production preparation time; parameter adjustment interval time and parameter adjustment time are all assumed to obey exponential distribution.

Images