Collection and treatment system and method for waste gas of lithium battery production workshop

Abstract

The invention discloses a lithium battery production workshop waste gas collection and treatment system and method, and relates to the technical field of industrial waste gas treatment.The method comprises the steps that historical operation data, adsorbent dosage and waste gas flow are collected, and a treatment efficiency prediction model is constructed and trained; constructing a cascade model between adjacent treatment units to form a treatment unit cascade model, obtaining input waste gas characteristics of each treatment unit, and defining treatment performance indexes; defining a state space, an action space and a reward function, constructing an optimization target of each processing unit, learning an optimal strategy of the optimization target, and taking the optimal strategy as a control strategy of the adaptive feedback controller; monitoring the operation state data in real time, predicting the real-time processing efficiency by using the processing efficiency prediction model, and calculating the fault degree; the fault type is classified into normal operation and abnormal operation, the fault-tolerant control strategy is constructed, and when the fault type is abnormal operation, the fault-tolerant control strategy is triggered according to the fault type, so that the economic benefit of waste gas treatment is improved.

Description

Technical Field

[0001] The present application relates to the technical field of industrial waste gas treatment, and particularly to a waste gas collection and treatment system and method for a lithium battery production workshop. Background Art

[0002] During the production process of lithium batteries, various harmful waste gases are generated, such as volatile organic compounds (VOCs), fluorides, acidic gases, etc. If these waste gases are directly discharged without effective treatment, they will pollute the environment and endanger the balance of the ecological system. At the same time, harmful waste gases also pose a threat to the physical health of workshop workers. Long-term exposure may lead to occupational diseases such as respiratory diseases and nervous system damage. Therefore, it is very necessary to efficiently and environmentally collect and treat the waste gas in the lithium battery production workshop. This is not only an embodiment of the enterprise's fulfillment of environmental protection responsibilities, but also an important measure to ensure the occupational health and safety of employees. Currently, the commonly used waste gas treatment methods in the lithium battery industry mainly include adsorption method, catalytic combustion method, plasma method, etc.

[0003] A Chinese patent with the authorization announcement number CN114191949B titled "A Method, System and Readable Storage Medium for Treating Waste Gas in Lithium Battery Production" includes: presetting m waste gas emission nodes and n - level waste gas treatment devices; respectively and real - time detecting the waste gas concentration values at each waste gas emission node through waste gas detection sensors; based on the waste gas concentration value at waste gas emission node A i , determining the falling concentration range and finding out the waste gas treatment device B j adapted to this concentration range; by controlling the solenoid valve at waste gas emission node A i , introducing the waste gas into the waste gas treatment device B j for treatment; if j = n, then monitoring whether the discharged gas from the waste gas treatment device B j meets the waste gas emission standard by a monitoring sensor. If so, it is allowed to be discharged. If not, the discharge is stopped. If j < n, then introducing the gas discharged from the waste gas treatment device B j into the waste gas treatment device B j+1 for the (j + 1) - level waste gas treatment.

[0004] The existing lithium battery waste gas treatment technologies such as adsorption method, catalytic combustion method, plasma method, etc. generally have problems of high energy consumption and high operating cost. How to reduce the energy consumption of the waste gas treatment system while ensuring the waste gas treatment effect, thereby reducing the operating cost of waste gas treatment, and further improving the economic and environmental benefits of lithium battery production enterprises has become an urgent problem to be solved. Summary of the Invention

[0005] The present application aims to solve at least one of the technical problems in the related art to a certain extent. To this end, one purpose of the present application is to propose a system and method for collecting and treating waste gas in a lithium battery production workshop, thereby improving the economic benefits of waste gas treatment.

[0006] One aspect of the present application provides a method for collecting and treating waste gas from a lithium battery production workshop, comprising:

[0007] Step S100: collecting historical operation data, adsorbent usage and exhaust gas flow of each processing unit, building a processing efficiency prediction model for each processing unit, and using the historical operation data to train the processing efficiency prediction model;

[0008] The input exhaust gas characteristics refer to the characteristic vectors of the exhaust gas entering each processing unit, which are obtained by online monitoring equipment and offline sampling and analysis;

[0009] Step S200: constructing a cascade model between adjacent processing units based on historical operation data, adsorbent dosage and exhaust gas flow, forming a processing unit cascade model, obtaining input exhaust gas characteristics of each processing unit, and defining processing performance indicators;

[0010] Step S300: define the state space, action space and reward function, construct the optimization target of each processing unit, learn the optimal strategy of the optimization target by training the Actor-Critic network, and use the optimal strategy as the control strategy of the adaptive feedback controller;

[0011] Step S400: monitor the operation status data of each processing unit in real time, predict the real-time processing efficiency of each processing unit using the processing efficiency prediction model, and calculate its fault degree;

[0012] Step S500: Classify the fault type into normal operation and abnormal operation according to the fault degree, construct a fault-tolerant control strategy for each fault type, and when the fault type is abnormal operation, trigger the corresponding fault-tolerant control strategy according to the fault type.

[0013] Preferably, the historical operation data, adsorbent dosage and exhaust gas flow of each processing unit are collected, and a processing efficiency prediction model is constructed for each processing unit. The specific method of using the historical operation data to train the processing efficiency prediction model is:

[0014] Step S110: Collect historical operation data, adsorbent usage and exhaust gas flow rate of the i-th processing unit, the historical operation data includes: input exhaust gas characteristics f i , process parameters p i and processing efficiency i , record the sampling time of each historical operation data;

[0015] Step S120: constructing an LSTM model, wherein the LSTM model includes an input layer, an LSTM layer, and a fully connected layer, using a sigmoid function as an activation function, and designing a loss function;

[0016] Step S130: Input exhaust gas characteristics f i , process parameters p i and processing efficiency r i Sort by time sequence, preset the prediction time step a1, window length a2 and sliding step a3, and use the sliding window method to input the exhaust gas feature f i , process parameters p i and processing efficiency i Divide into N xl training samples, and the input exhaust gas feature f i , process parameters p i As input data, the processing efficiency of the predicted time step a1 is taken as output data, and the accurate prediction of the actual processing efficiency is taken as the prediction target. When the value of the loss function converges to the minimum, the training is completed and the processing efficiency prediction model is obtained.

[0017] Preferably, the cascade model between adjacent treatment units is constructed based on historical operation data, adsorbent dosage and exhaust gas flow rate to form a treatment unit cascade model, and the input exhaust gas characteristics of each treatment unit are obtained. The specific method for defining the treatment performance index is:

[0018] Step S210: construct a cascade model between adjacent processing units. The cascade model G i The function expression is: Among them, f i+1 is the input exhaust gas characteristic of the i+1th processing unit, K i is the adsorption equilibrium constant of the ith treatment unit, m i is the amount of adsorbent used in the ith treatment unit, Q i is the exhaust gas flow rate of the i-th treatment unit, i+1∈{1,2,...,N}, N is the total number of treatment units;

[0019] Step S220: Connect the cascade models between adjacent processing units in the order of the processing units to form a processing unit cascade model {f2, f3, ..., f N}, the input exhaust gas characteristics of each treatment unit are obtained through the treatment unit cascade model;

[0020] Step S230: defining a processing performance index of each processing unit, predicting the processing efficiency of each processing unit through a processing efficiency prediction model, obtaining processing energy consumption through direct measurement statistics, and obtaining the processing performance index by weighted summation of processing efficiency and processing energy consumption;

[0021] The calculation formula of the processing performance index is: zb =ω i × i -α×E i , where ω i and α are the weight coefficients of processing efficiency and processing energy consumption, respectively. i is the processing energy consumption of the i-th processing unit.

[0022] Preferably, the specific method of defining the state space, action space and reward function, constructing the optimization target of each processing unit, learning the optimal strategy of the optimization target by training the Actor-Critic network, and using the optimal strategy as the control strategy of the adaptive feedback controller is:

[0023] Step S310: define the state space of reinforcement learning, where the state of the i-th processing unit at sampling time t is s i,t =[f i,t ,r i,t ,t], where f i,t is the input exhaust gas characteristic of the i-th processing unit at sampling time t, r i,t is the processing efficiency of the i-th processing unit at sampling time t predicted by the processing efficiency prediction model;

[0024] Define the action space of reinforcement learning, where the action of the i-th processing unit at sampling time t is a i,t =p i,t , where p i,t is the process parameter of the i-th processing unit at sampling time t;

[0025] Define the reward function of reinforcement learning, take the processing performance index as the reward value of the reward function, and the reward value of the i-th processing unit at sampling time t is R i,t =ω i × i,t -α i ×E i,t , where ω i , α i are the efficiency weight and energy consumption weight of the i-th processing unit, E i,t is the processing energy consumption of the i-th processing unit at sampling time t;

[0026] Step S320: For each processing unit i, construct an optimization target, the expression of which is: max E[∑γ t ×R i,t ], where γ is the discount factor and E[·] represents the expected value;

[0027] Step S330: Construct an Actor-Critic network. The Actor network is in state s i,t is input, output action a i,t , the Critic network is in state s i,t and action a i,t is the input and output state-action value function SA(s i,t ,a i,t ), the Actor network and the Critic network adopt a multi-layer feedforward neural network structure, and the ReLU function is selected as the activation function;

[0028] Step S340: The Actor network is based on the state s i,t Generate action a i,t , and the action a i,t Applied to the i-th processing unit, collect the state s at the next sampling time i,t+1 and the reward value R at the current sampling moment i,t , using the optimization objective to learn the optimal strategy Π i,opt , using the optimal strategy Π i,opt For action a i,t Perform importance sampling to get weighted actions Using weighted actions and the reward value R at the current sampling moment i,t Calculate the loss function of the Critic network, and use the Adam optimizer and back-propagation algorithm to optimize the parameters of the Actor network and the Critic network;

[0029] Step S350: When the reward value reaches the reward stability value R, the training of the Actor-Critic network is completed. The Actor-Critic network updates the action at the next moment according to the current action, state and reward value, and uses the optimal strategy Π learned as the optimization target i,opt ;

[0030] Step S360: For each processing unit, construct an adaptive feedback controller Ci, and learn the optimal strategy Π of the optimization target of the i-th processing unit i,opt As the control strategy of the adaptive feedback controller Ci.

[0031] Preferably, the specific method of real-time monitoring the operating status data of each processing unit, using the processing efficiency prediction model to predict the real-time processing efficiency of each processing unit, and calculating its fault degree is:

[0032] Step S410: monitor the operating status of each processing unit in real time and obtain its operating status data, wherein the operating status data includes real-time exhaust gas characteristics, real-time processing energy consumption, real-time adsorbent usage, and real-time exhaust gas flow;

[0033] Step S420: predicting the current real-time processing efficiency using the processing efficiency prediction model;

[0034] Step S430: Calculating the fault degree of each processing unit according to the operation status data and the real-time processing efficiency;

[0035] The calculation formula of the fault degree is: Among them, FD i,t is the fault degree of the i-th processing unit at sampling time t, η i,opt 、E i,opt ,q i,opt 、F i,opt are the optimal treatment efficiency, optimal treatment energy consumption, optimal adsorbent dosage and optimal exhaust gas flow rate of the i-th treatment unit, respectively, i,t 、E i,t ,q i,t 、F i,t They represent the real-time processing efficiency, real-time processing energy consumption, real-time adsorbent usage and real-time exhaust gas flow of the i-th processing unit at sampling time t, ω η ,ω E ,ω q ,ω F They are the weight coefficients of treatment efficiency, treatment energy consumption, adsorbent dosage and waste gas flow rate.

[0036] Preferably, the fault type is classified into normal operation and abnormal operation according to the fault degree, and a fault-tolerant control strategy is constructed for each fault type. When the fault type is abnormal operation, the specific method of triggering the corresponding fault-tolerant control strategy according to the fault type is:

[0037] Step S510: Classify the fault type into normal operation and abnormal operation according to the fault degree. The abnormal operation includes a slight fault, a moderate fault and a severe fault. When the fault degree is within the normal range [0, FD minor ), the fault type is normal operation; when the fault degree is in the mild range [FD minor ,FD moderate ), the fault type is a mild fault; when the fault level is in the moderate range [FD moderate ,FD severe ), the fault type is moderate fault; when the fault level is in the severe range [FD severe ,1], the fault type is severe fault;

[0038] Step S520: construct a fault-tolerant control strategy for each fault type. When the fault type is normal operation, maintain the optimal strategy unchanged; when the fault type is a minor fault, increase the real-time adsorbent dosage by δ1 on the basis of the optimal strategy; when the fault type is a moderate fault, increase the real-time adsorbent dosage by δ1 and reduce the real-time exhaust gas flow by δ2 on the basis of the optimal strategy; when the fault type is a major fault, stop inputting exhaust gas to the i-th processing unit and enable the standby processing unit;

[0039] Step S530: confirm the fault type of the current i-th processing unit according to the fault degree, and trigger the corresponding fault-tolerant control strategy.

[0040] Preferably, the normal range is divided into: defining a normal operation threshold Among them, η min is the minimum value of processing efficiency, E max ,q max 、F max are the maximum values ​​of treatment energy consumption, adsorbent dosage and waste gas flow rate, respectively, opt 、E opt ,q opt 、F opt are the values ​​of treatment efficiency, treatment energy consumption, adsorbent dosage and exhaust gas flow rate under the optimal operating state;

[0042] Define Mild Threshold FD minor , and get the normal range [0,FD minor ), the calculation formula of the mild threshold is: FD minor =1.5×FD normal ;

[0043] The method for dividing the mild range is as follows: define the moderate threshold FD moderate , get the mild range [FD minor ,FD moderate ), the calculation formula of the moderate threshold is: FD moderate =2×FD normal ;

[0044] The method of dividing the moderate range is as follows: define the severe threshold FD severe , and obtain the medium range [FD moderate ,FD severe ), the calculation formula of the severe threshold is: FD severe =3×FD normal ;

[0045] The method for dividing the severe range is as follows: the severe range [FD severe ,1].

[0046] One aspect of the present application provides a system for collecting and treating waste gas from a lithium battery production workshop, comprising:

[0047] The processing efficiency prediction module is used to collect the historical operation data, adsorbent usage and exhaust gas flow of each processing unit, build a processing efficiency prediction model for each processing unit, and train the processing efficiency prediction model using the historical operation data;

[0048] Cascade model building module, used to build a cascade model between adjacent treatment units based on historical operation data, adsorbent dosage and exhaust gas flow, form a treatment unit cascade model, obtain the input exhaust gas characteristics of each treatment unit, and define treatment performance indicators;

[0049] The optimal strategy learning module is used to define the state space, action space and reward function, build the optimization target of each processing unit, learn the optimal strategy of the optimization target by training the Actor-Critic network, and use the optimal strategy as the control strategy of the adaptive feedback controller;

[0050] A fault degree calculation module is used to monitor the operating status data of each processing unit in real time, predict the real-time processing efficiency of each processing unit using a processing efficiency prediction model, and calculate its fault degree;

[0051] The fault-tolerant control trigger module is used to classify the fault type into normal operation and abnormal operation according to the fault degree, and to build a fault-tolerant control strategy for each fault type. When the fault type is abnormal operation, the corresponding fault-tolerant control strategy is triggered according to the fault type.

[0052] One aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for collecting and treating waste gas in a lithium battery production workshop are implemented.

[0053] One aspect of the present application provides a readable storage medium storing a computer program suitable for loading by a processor to execute the steps in a method for collecting and treating waste gas in a lithium battery production workshop.

[0054] The system and method for collecting and treating waste gas from a lithium battery production workshop proposed in this application have the following advantages over the prior art:

[0055] This application constructs a treatment efficiency prediction model and a cascade model, and uses a data-driven modeling method to fully explore the implicit rules in the historical data of exhaust gas treatment. The cascade model makes the various treatment units in the system no longer isolated, but connected into an organic whole, reflecting the overall performance of the exhaust gas treatment system.

[0056] This application uses the Actor-Critic algorithm in reinforcement learning to learn the optimal control strategy through continuous interaction with the environment, and use it to guide the actions of the adaptive controller. This end-to-end learning optimization method can continuously improve the strategy based on the actual feedback of the system, find the global optimal control solution, and improve the economic benefits of exhaust gas treatment from the perspective of processing energy consumption and processing efficiency.

[0057] This application adds the functions of fault diagnosis and fault-tolerant control, further improving the safety and reliability of the system; this control approach of active prevention, timely diagnosis, and automatic fault tolerance avoids the occurrence and spread of accidents, reduces dependence on human intervention, and improves the economic benefits of exhaust gas treatment from the perspective of fault prevention. It is an important manifestation of moving towards an autonomous intelligent system.

[0058] This application integrates data-driven, intelligent optimization, fault diagnosis, and multi-objective trade-off methods, which greatly improves the performance and intelligence level of the exhaust gas treatment system, and is of great significance for energy conservation, emission reduction, and environmental protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A flow chart of the method for collecting and treating waste gas from a lithium battery production workshop provided in this application;

[0060] Figure 2 Functional module diagram of the waste gas collection and treatment system for the lithium battery production workshop provided in this application;

[0061] Figure 3 A schematic diagram of the structure of an electronic device provided in this application;

[0062] Figure 4 It is a schematic diagram of the structure of a readable storage medium provided by this application. DETAILED DESCRIPTION

[0063] In order to better understand the present application, various aspects of the present application will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are only descriptions of exemplary embodiments of the present application and do not limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0064] In the accompanying drawings, the size, dimensions and shapes of the elements have been slightly adjusted for ease of illustration. The drawings are for illustration only and are not strictly drawn to scale. As used herein, the terms "substantially", "approximately" and similar terms are used as terms of approximation, not as terms of degree, and are intended to illustrate the inherent deviations in measurements or calculations that will be recognized by those of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these processes occur in actual operation, unless otherwise specified or can be derived from the context.

[0065] It should also be understood that expressions such as "including", "comprising", "having", "containing" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0066] Unless otherwise specified, all words (including engineering terms and scientific and technological terms) used in this article have the same meaning as those commonly understood by ordinary technicians in the field to which this application belongs. It should also be understood that, unless clearly stated in this application, words defined in common dictionaries should be interpreted as having the same meaning as their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0067] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0068] Example 1

[0069] like Figure 1 As shown, the method for collecting and treating waste gas from a lithium battery production workshop provided in this application includes:

[0070] Step S100: collecting historical operation data, adsorbent usage and exhaust gas flow of each processing unit, building a processing efficiency prediction model for each processing unit, and using the historical operation data to train the processing efficiency prediction model;

[0071] The input exhaust gas characteristics refer to the characteristic vectors of the exhaust gas entering each processing unit, which are obtained by online monitoring equipment and offline sampling and analysis;

[0072] The specific method of collecting the historical operation data, adsorbent usage and exhaust gas flow of each processing unit, building a processing efficiency prediction model for each processing unit, and using the historical operation data to train the processing efficiency prediction model is as follows:

[0073] Step S110: Collect historical operation data, adsorbent usage and exhaust gas flow rate of the i-th processing unit, the historical operation data includes: input exhaust gas characteristics f i , process parameters p i and processing efficiency i , record the sampling time of each historical operation data;

[0074] Step S120: constructing an LSTM model, wherein the LSTM model includes an input layer, an LSTM layer, and a fully connected layer, using a sigmoid function as an activation function, and designing a loss function;

[0075] The calculation formula of the loss function is: in, represents the processing efficiency of the prediction, r i represents the actual processing efficiency, λ L2 represents the regularization strength coefficient, θ M Represents the model parameters of the LSTM model, ||·|| is the symbol for calculating the modulus length;

[0076] The regularization strength coefficient is used to control overfitting and is set by those skilled in the art according to the balance between mean square error and L2 regularization;

[0077] Step S130: Input exhaust gas characteristics f i , process parameters p i and processing efficiency i Sort by time sequence, preset the prediction time step a1, window length a2 and sliding step a3, and use the sliding window method to input the exhaust gas feature f i , process parameters p i and processing efficiency i Divide into N xl training samples, and the input exhaust gas feature f i , process parameters p i As input data, the processing efficiency of the predicted time step a1 is used as output data, and the accurate prediction of the actual processing efficiency is used as the prediction target. When the value of the loss function converges to the minimum, the training is completed and the processing efficiency prediction model is obtained;

[0078] The prediction time step a1, window length a2 and sliding step a3 are preset by those skilled in the art according to the requirements of model training.

[0079] Step S200: constructing a cascade model between adjacent processing units based on historical operation data, adsorbent dosage and exhaust gas flow, forming a processing unit cascade model, obtaining input exhaust gas characteristics of each processing unit, and defining processing performance indicators;

[0080] The specific method of constructing a cascade model between adjacent treatment units based on historical operation data, adsorbent dosage and exhaust gas flow, forming a treatment unit cascade model, obtaining input exhaust gas characteristics of each treatment unit, and defining treatment performance indicators is as follows:

[0081] Step S210: construct a cascade model between adjacent processing units. The cascade model G i The function expression is: Among them, f i+1 is the input exhaust gas characteristic of the i+1th processing unit, K i is the adsorption equilibrium constant of the ith treatment unit, m i is the amount of adsorbent used in the ith treatment unit, Q i is the exhaust gas flow rate of the i-th treatment unit, i+1∈{1,2,...,N}, N is the total number of treatment units;

[0082] The adsorption equilibrium constant K i The setting method is as follows: for each adjacent treatment unit (i, i+1), the input exhaust gas characteristics, adsorbent dosage and exhaust gas flow rate of the i-th treatment unit are taken as input, and the input exhaust gas characteristics of the i+1-th treatment unit are taken as output, and the cascade model is fitted to obtain the value of the adsorption equilibrium constant of the i-th treatment unit;

[0083] Step S220: Connect the cascade models between adjacent processing units in the order of the processing units to form a processing unit cascade model {f2, f3, ..., f N}, the input exhaust gas characteristics of each treatment unit are obtained through the treatment unit cascade model;

[0084] Step S230: defining a processing performance index of each processing unit, predicting the processing efficiency of each processing unit through a processing efficiency prediction model, obtaining processing energy consumption through direct measurement statistics, and obtaining the processing performance index by weighted summation of processing efficiency and processing energy consumption;

[0085] The calculation formula of the processing performance index is: zb =ω i × i -α×E i , where ω i and α are the weight coefficients of processing efficiency and processing energy consumption, respectively. i is the processing energy consumption of the i-th processing unit;

[0086] The weight coefficients of the processing efficiency and processing energy consumption are set by those skilled in the art according to requirements.

[0087] Step S300: define the state space, action space and reward function, construct the optimization target of each processing unit, learn the optimal strategy of the optimization target by training the Actor-Critic network, and use the optimal strategy as the control strategy of the adaptive feedback controller;

[0088] The specific method of defining the state space, action space and reward function, constructing the optimization target of each processing unit, learning the optimal strategy of the optimization target by training the Actor-Critic network, and using the optimal strategy as the control strategy of the adaptive feedback controller is as follows:

[0089] Step S310: define the state space, action space and reward function of reinforcement learning;

[0090] The specific method for defining the state space, action space and reward function of reinforcement learning is:

[0091] Step S311: define the state space of reinforcement learning, where the state of the i-th processing unit at sampling time t is s i,t =[f i,t ,r i,t ,t], where f i,t is the input exhaust gas characteristic of the i-th processing unit at sampling time t, r i,t is the processing efficiency of the i-th processing unit at sampling time t predicted by the processing efficiency prediction model;

[0092] Step S312: define the action space of reinforcement learning, where the action of the i-th processing unit at sampling time t is a i,t =p i,t , where p i,t is the process parameter of the i-th processing unit at sampling time t;

[0093] Step S313: Define the reward function of reinforcement learning, use the processing performance index as the reward value of the reward function, and the reward value of the i-th processing unit at sampling time t is R i,t =ω i × i,t -α i ×E i,t , where ω i , α i are the efficiency weight and energy consumption weight of the i-th processing unit, E i,t is the processing energy consumption of the i-th processing unit at sampling time t;

[0094] Step S320: For each processing unit i, construct an optimization target, the expression of which is: max E[∑γ t ×R i,t ], where γ is the discount factor and E[·] represents the expected value;

[0095] γ is set by technicians in this field as needed to balance the importance of the current reward value and the reward value at the next sampling moment. The closer γ is to 1, the more importance is attached to future rewards; the closer γ is to 0, the more importance is attached to immediate rewards.

[0096] Step S330: Construct an Actor-Critic network. The Actor network is in state s i,t is input, output action a i,t , the Critic network is in state s i,t and action a i,t is the input and output state-action value function SA(s i,t ,a i,t ), the Actor network and the Critic network adopt a multi-layer feedforward neural network structure, and the ReLU function is selected as the activation function;

[0097] Step S340: The Actor network is based on the state s i,t Generate action a i,t , and the action a i,t Applied to the i-th processing unit, collect the state s at the next sampling time i,t+1 and the reward value R at the current sampling moment i,t , using the optimization objective to learn the optimal strategy Π i,opt , using the optimal strategy Π i,opt For action a i,t Perform importance sampling to get weighted actions Using weighted actions and the reward value R at the current sampling moment i,t Calculate the loss function of the Critic network, and use the Adam optimizer and back-propagation algorithm to optimize the parameters of the Actor network and the Critic network;

[0098] The calculation formula of the loss function of the Critic network is: L SA =SA(s i,t ,a i,t )-(R i,t +γ×max a′ SA(s i,t+1 ,a')), where L SA is the loss function of the Critic network, max a′ SA(s i,t+1,a') represents the state s at the next sampling time t+1 i,t+1 The maximum state-action value function value of all actions a' under

[0099] Maximum state-action value function value max a′ SA(s i,t+1 ,a') means in state s i,t+1 The expected reward of the best action to take under

[0100] Step S350: When the reward value reaches the reward stability value R, the training of the Actor-Critic network is completed. The Actor-Critic network updates the action at the next moment according to the current action, state and reward value, and uses the optimal strategy Π learned as the optimization target i,opt ;

[0101] The reward stability value R is set by those skilled in the art according to actual needs;

[0102] When the reward function converges to the reward stability value R within the preset number of steps, and the upward or downward trend is less than or equal to β, it means that the training of the Actor-Critic network has been completed;

[0103] The value of β is set by those skilled in the art according to actual needs;

[0104] Step S360: For each processing unit, construct an adaptive feedback controller Ci, and learn the optimal strategy Π of the optimization target of the i-th processing unit i,opt As the control strategy of the adaptive feedback controller Ci.

[0105] Step S400: monitor the operation status data of each processing unit in real time, predict the real-time processing efficiency of each processing unit using the processing efficiency prediction model, and calculate its fault degree;

[0106] The specific method of real-time monitoring the operating status data of each processing unit, using the processing efficiency prediction model to predict the real-time processing efficiency of each processing unit, and calculating its fault degree is:

[0107] Step S410: monitor the operating status of each processing unit in real time and obtain its operating status data, wherein the operating status data includes real-time exhaust gas characteristics, real-time processing energy consumption, real-time adsorbent usage, and real-time exhaust gas flow;

[0108] Step S420: predicting the current real-time processing efficiency using the processing efficiency prediction model;

[0109] Step S430: Calculating the fault degree of each processing unit according to the operation status data and the real-time processing efficiency;

[0110] The calculation formula of the fault degree is: Among them, FD i,t is the fault degree of the i-th processing unit at sampling time t, η i,opt 、E i,opt ,q i,opt 、F i,opt are the optimal treatment efficiency, optimal treatment energy consumption, optimal adsorbent dosage and optimal exhaust gas flow rate of the i-th treatment unit, respectively, i,t 、E i,t ,q i,t 、F i,t They represent the real-time processing efficiency, real-time processing energy consumption, real-time adsorbent usage and real-time exhaust gas flow of the i-th processing unit at sampling time t, ω η ,ω E ,ω q ,ω F are the weight coefficients of treatment efficiency, treatment energy consumption, adsorbent dosage and waste gas flow rate;

[0111] The weight coefficients of the treatment efficiency, treatment energy consumption, adsorbent dosage and exhaust gas flow rate are used to balance the impact of different operating status data on the degree of fault, and are set by technicians in this field according to actual needs.

[0112] Step S500: classify the fault type into normal operation and abnormal operation according to the fault degree, construct a fault-tolerant control strategy for each fault type, and when the fault type is abnormal operation, trigger the corresponding fault-tolerant control strategy according to the fault type;

[0113] The fault type is classified into normal operation and abnormal operation according to the fault degree, and a fault-tolerant control strategy is constructed for each fault type. When the fault type is abnormal operation, the specific method of triggering the corresponding fault-tolerant control strategy according to the fault type is as follows:

[0114] Step S510: Classify the fault type into normal operation and abnormal operation according to the fault degree. The abnormal operation includes a slight fault, a moderate fault and a severe fault. When the fault degree is within the normal range [0, FD minor ), the fault type is normal operation; when the fault degree is in the mild range [FD minor ,FD moderate ), the fault type is a mild fault; when the fault level is in the moderate range [FD moderate ,FD severe ), the fault type is moderate fault; when the fault level is in the severe range [FD severe ,1], the fault type is severe fault;

[0115] The normal range is divided as follows: define the normal operation threshold FD normal = Among them, η min is the minimum value of processing efficiency, E max ,q max 、F max are the maximum values ​​of treatment energy consumption, adsorbent dosage and waste gas flow rate, respectively, opt 、E opt ,q opt 、F opt are the values ​​of treatment efficiency, treatment energy consumption, adsorbent dosage and exhaust gas flow rate under the optimal operating state;

[0118] The η min 、E max ,q max 、F max , η opt 、E opt ,q opt 、F opt The value of is obtained from the data collected by the Actor-Critic network during actual operation;

[0119] Define Mild Threshold FD minor , and get the normal range [0,FD minor ), the calculation formula of the mild threshold is: FD minor =1.5×FD normal ;

[0120] The method for dividing the mild range is as follows: define the moderate threshold FD moderate , get the mild range [FD minor ,FD moderate ), the calculation formula of the moderate threshold is: FD moderate =2×FD normal ;

[0121] The method of dividing the moderate range is as follows: define the severe threshold FD severe , and obtain the medium range [FD moderate ,FD severe ), the calculation formula of the severe threshold is: FD severe =3×FD normal ;

[0122] The method for dividing the severe range is as follows: the severe range [FD severe ,1];

[0123] Step S520: construct a fault-tolerant control strategy for each fault type. When the fault type is normal operation, maintain the optimal strategy unchanged; when the fault type is a minor fault, increase the real-time adsorbent dosage by δ1 on the basis of the optimal strategy; when the fault type is a moderate fault, increase the real-time adsorbent dosage by δ1 and reduce the real-time exhaust gas flow by δ2 on the basis of the optimal strategy; when the fault type is a major fault, stop inputting exhaust gas to the i-th processing unit and enable the standby processing unit;

[0124] The values ​​of the real-time adsorbent dosage increase by δ1 and the real-time exhaust gas flow reduction by δ2 are set by those skilled in the art according to actual needs;

[0125] Step S530: confirming the fault type of the current i-th processing unit according to the fault degree, and triggering a corresponding fault-tolerant control strategy;

[0126] When the system is normal, the optimized target control strategy is run to save energy and increase efficiency while taking into account the stability of the system to avoid malfunctions caused by severe fluctuations in operating conditions;

[0127] Once a fault occurs, switch to the fault-tolerant control strategy, adopt the corresponding strategy, and resume the optimization target control after the fault is eliminated.

[0128] Example 2

[0129] like Figure 2 As shown, the waste gas collection and treatment system for the lithium battery production workshop provided in this application includes:

[0130] The processing efficiency prediction module is used to collect the historical operation data, adsorbent usage and exhaust gas flow of each processing unit, build a processing efficiency prediction model for each processing unit, and train the processing efficiency prediction model using the historical operation data;

[0131] Cascade model building module, used to build a cascade model between adjacent treatment units based on historical operation data, adsorbent dosage and exhaust gas flow, form a treatment unit cascade model, obtain the input exhaust gas characteristics of each treatment unit, and define treatment performance indicators;

[0132] The optimal strategy learning module is used to define the state space, action space and reward function, build the optimization target of each processing unit, learn the optimal strategy of the optimization target by training the Actor-Critic network, and use the optimal strategy as the control strategy of the adaptive feedback controller;

[0133] A fault degree calculation module is used to monitor the operating status data of each processing unit in real time, predict the real-time processing efficiency of each processing unit using a processing efficiency prediction model, and calculate its fault degree;

[0134] The fault-tolerant control trigger module is used to classify the fault type into normal operation and abnormal operation according to the fault degree, and to build a fault-tolerant control strategy for each fault type. When the fault type is abnormal operation, the corresponding fault-tolerant control strategy is triggered according to the fault type.

[0135] Example 3

[0136] Figure 3 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 3 As shown, according to another aspect of the present application, an electronic device is also provided. The electronic device may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are run by one or more processors, the method for collecting and treating waste gas in a lithium battery production workshop as described above may be executed.

[0137] The method or system according to the embodiment of the present application can also be used by Figure 3 The electronic device architecture shown in FIG. Figure 3 As shown, the electronic device may include a bus, one or more CPUs, a read-only memory (ROM), a random access memory (RAM), a communication port connected to a network, an input / output component, a hard disk, etc. A storage device in the electronic device, such as a ROM or a hard disk, may store the method for collecting and treating waste gas from a lithium battery production workshop provided in this application. The method for collecting and treating waste gas in a lithium battery production workshop may, for example, include: collecting historical operating data, adsorbent dosage and waste gas flow of each processing unit, constructing a processing efficiency prediction model for each processing unit, and using historical operating data to train the processing efficiency prediction model; constructing a cascade model between adjacent processing units based on historical operating data, adsorbent dosage and waste gas flow, forming a processing unit cascade model, obtaining the input waste gas characteristics of each processing unit, and defining processing performance indicators; defining state space, action space and reward function, constructing the optimization target of each processing unit, and learning the optimal strategy of the optimization target by training the Actor-Critic network, and using the optimal strategy as the control strategy of the adaptive feedback controller; real-time monitoring of the operating status data of each processing unit, using the processing efficiency prediction model to predict the real-time processing efficiency of each processing unit, and calculating its degree of failure; classifying the fault type into normal operation and abnormal operation according to the degree of failure, constructing a fault-tolerant control strategy for each fault type, and when the fault type is abnormal operation, triggering the corresponding fault-tolerant control strategy according to the fault type. Furthermore, the electronic device may also include a user interface. Of course, Figure 3 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 3 One or more components of an electronic device are shown.

[0138] Example 4

[0139] Figure 4 Schematic diagram of a readable storage medium structure provided by an embodiment of the present application. Figure 4 As shown, it is a readable storage medium according to an embodiment of the present application. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by the processor, the method for collecting and treating waste gas from a lithium battery production workshop according to an embodiment of the present application described with reference to the above figures can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0140] In addition, according to an embodiment of the present application, the process described above with reference to the flowchart may be implemented as a computer software program. For example, the present application provides a non-temporary machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application, for example: collecting historical operating data, adsorbent usage and exhaust gas flow of each processing unit, constructing a processing efficiency prediction model for each processing unit, and training the processing efficiency prediction model using historical operating data; constructing a cascade model between adjacent processing units based on historical operating data, adsorbent usage and exhaust gas flow, forming a processing unit cascade model, obtaining the input exhaust gas characteristics of each processing unit, and defining processing performance indicators; defining a state space, an action space and a reward function, constructing an optimization target for each processing unit, learning the optimal strategy of the optimization target by training an Actor-Critic network, and using the optimal strategy as the control strategy of an adaptive feedback controller; monitoring the operating status data of each processing unit in real time, predicting the real-time processing efficiency of each processing unit using a processing efficiency prediction model, and calculating its fault degree; classifying the fault type into normal operation and abnormal operation according to the fault degree, constructing a fault-tolerant control strategy for each fault type, and when the fault type is abnormal operation, triggering the corresponding fault-tolerant control strategy according to the fault type. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are performed.

[0141] The methods, apparatuses, and devices of the present application may be implemented in many ways. For example, the methods, apparatuses, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers recording media storing programs for executing the method according to the present application.

[0142] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0143] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for collecting and treating waste gas from a lithium battery production workshop, characterized in that: include: Collect historical operation data, adsorbent usage and exhaust gas flow of each treatment unit, build a treatment efficiency prediction model for each treatment unit, and use the historical operation data to train the treatment efficiency prediction model; Based on historical operation data, adsorbent dosage and exhaust gas flow, a cascade model between adjacent treatment units is constructed to form a treatment unit cascade model, obtain the input exhaust gas characteristics of each treatment unit, and define the treatment performance index; Define the state space, action space and reward function, construct the optimization target of each processing unit, learn the optimal strategy of the optimization target by training the Actor-Critic network, and use the optimal strategy as the control strategy of the adaptive feedback controller; Monitor the operating status data of each processing unit in real time, use the processing efficiency prediction model to predict the real-time processing efficiency of each processing unit, and calculate its fault degree; The fault types are classified into normal operation and abnormal operation according to the fault severity, and a fault-tolerant control strategy is constructed for each fault type. When the fault type is abnormal operation, the corresponding fault-tolerant control strategy is triggered according to the fault type.

2. The method for collecting and treating waste gas from a lithium battery production workshop according to claim 1, characterized in that: The specific method of collecting the historical operation data, adsorbent usage and exhaust gas flow of each processing unit, building a processing efficiency prediction model for each processing unit, and using the historical operation data to train the processing efficiency prediction model is as follows: Collect historical operation data, adsorbent usage and exhaust gas flow of the i-th processing unit, the historical operation data includes: input exhaust gas characteristics f i , process parameters p i and processing efficiency i , record the sampling time of each historical operation data; Construct an LSTM model, which includes an input layer, an LSTM layer, and a fully connected layer, uses a sigmoid function as an activation function, and designs a loss function; The exhaust gas characteristic f i , process parameters p i and processing efficiency i Sort by time sequence, preset the prediction time step a1, window length a2 and sliding step a3, and use the sliding window method to input the exhaust gas feature f i , process parameters p i and processing efficiency i Divide into N xl training samples, and the input exhaust gas feature f i , process parameters p i As input data, the processing efficiency of the predicted time step a1 is taken as output data, and the accurate prediction of the actual processing efficiency is taken as the prediction target. When the value of the loss function converges to the minimum, the training is completed and the processing efficiency prediction model is obtained.

3. The method for collecting and treating waste gas from a lithium battery production workshop according to claim 2, characterized in that: The specific method of constructing a cascade model between adjacent treatment units based on historical operation data, adsorbent dosage and exhaust gas flow, forming a treatment unit cascade model, obtaining input exhaust gas characteristics of each treatment unit, and defining treatment performance indicators is as follows: Construct a cascade model between adjacent processing units, the cascade model G i The function expression is: Among them, f i+1 is the input exhaust gas characteristic of the i+1th processing unit, K i is the adsorption equilibrium constant of the ith treatment unit, m i is the amount of adsorbent used in the ith treatment unit, Q i is the exhaust gas flow rate of the i-th treatment unit, i+1∈{1,2,...,N}, N is the total number of treatment units; The cascade models between adjacent processing units are connected in sequence according to the order of the processing units to form a processing unit cascade model {f2,f3,...,f N }, the input exhaust gas characteristics of each treatment unit are obtained through the treatment unit cascade model; Define the processing performance index of each processing unit, predict the processing efficiency of each processing unit through the processing efficiency prediction model, obtain the processing energy consumption through direct measurement statistics, and the processing performance index is obtained by the weighted sum of the processing efficiency and the processing energy consumption; The calculation formula of the processing performance index is: zb =ω i × i -α×E i , where ω i and α are the weight coefficients of processing efficiency and processing energy consumption, respectively. i is the processing energy consumption of the i-th processing unit.

4. The method for collecting and treating waste gas from a lithium battery production workshop as claimed in claim 3, characterized in that: The specific method of defining the state space, action space and reward function, constructing the optimization target of each processing unit, learning the optimal strategy of the optimization target by training the Actor-Critic network, and using the optimal strategy as the control strategy of the adaptive feedback controller is as follows: Define the state space of reinforcement learning. The state of the i-th processing unit in the state space at sampling time t is s i,t =[f i,t ,r i,t ,t], where f i,t is the input exhaust gas characteristic of the i-th processing unit at sampling time t, r i,t is the processing efficiency of the i-th processing unit at sampling time t predicted by the processing efficiency prediction model; Define the action space of reinforcement learning, where the action of the i-th processing unit at sampling time t is a i,t =p i,t , where p i,t is the process parameter of the i-th processing unit at sampling time t; Define the reward function of reinforcement learning, take the processing performance index as the reward value of the reward function, and the reward value of the i-th processing unit at sampling time t is R i,t =ω i × i,t -α i ×E i,t , where ω i , α i are the efficiency weight and energy consumption weight of the i-th processing unit, E i,t is the processing energy consumption of the i-th processing unit at sampling time t; For each processing unit i, an optimization target is constructed, and the expression of the optimization target is: max E[∑γ t ×R i,t ], where γ is the discount factor and E[·] represents the expected value; Construct an Actor-Critic network, where the Actor network is based on the state s i,t is input, output action a i,t , the Critic network is in state s i,t and action a i,t is the input and output state-action value function SA(s i,t ,a i,t ), the Actor network and the Critic network adopt a multi-layer feedforward neural network structure, and the ReLU function is selected as the activation function; Actor network according to the state s i,t Generate action a i,t , and the action a i,t Applied to the i-th processing unit, collect the state s at the next sampling time i,t+1 and the reward value R at the current sampling moment i,t , using the optimization objective to learn the optimal strategy Π i,opt , using the optimal strategy Π i,opt For action a i,t Perform importance sampling to get weighted actions Using weighted actions and the reward value R at the current sampling moment i,t Calculate the loss function of the Critic network, and use the Adam optimizer and back-propagation algorithm to optimize the parameters of the Actor network and the Critic network; When the reward value reaches the reward stability value R, the training of the Actor-Critic network is completed. The Actor-Critic network updates the action at the next moment according to the current action, state and reward value, and uses the optimal strategy Π learned as the optimization target i,opt ; For each processing unit, an adaptive feedback controller Ci is constructed to learn the optimal strategy Π of the optimization objective of the i-th processing unit. i,opt As the control strategy of the adaptive feedback controller Ci.

5. The method for collecting and treating waste gas from a lithium battery production workshop as claimed in claim 4, characterized in that: The specific method of real-time monitoring the operating status data of each processing unit, using the processing efficiency prediction model to predict the real-time processing efficiency of each processing unit, and calculating its fault degree is: Monitor the operating status of each processing unit in real time and obtain its operating status data, wherein the operating status data includes real-time exhaust gas characteristics, real-time processing energy consumption, real-time adsorbent usage, and real-time exhaust gas flow; Using the processing efficiency prediction model to predict the current real-time processing efficiency; Calculate the failure degree of each processing unit based on the operating status data and real-time processing efficiency; The calculation formula of the fault degree is: Among them, FD i,t is the fault degree of the i-th processing unit at sampling time t, η i,opt 、E i,opt ,q i,opt 、F i,opt are the optimal treatment efficiency, optimal treatment energy consumption, optimal adsorbent dosage and optimal exhaust gas flow rate of the i-th treatment unit, respectively, i,t 、E i,t ,q i,t 、F i,t They represent the real-time processing efficiency, real-time processing energy consumption, real-time adsorbent usage and real-time exhaust gas flow of the i-th processing unit at sampling time t, ω η ,ω E ,ω q ,ω F They are the weight coefficients of treatment efficiency, treatment energy consumption, adsorbent dosage and waste gas flow rate.

6. The method for collecting and treating waste gas from a lithium battery production workshop according to claim 5, characterized in that: The fault type is classified into normal operation and abnormal operation according to the fault degree, and a fault-tolerant control strategy is constructed for each fault type. When the fault type is abnormal operation, the specific method of triggering the corresponding fault-tolerant control strategy according to the fault type is as follows: According to the fault severity, the fault types are divided into normal operation and abnormal operation. Abnormal operation includes mild fault, moderate fault and severe fault. When the fault severity is within the normal range [0, FD minor ), the fault type is normal operation; when the fault degree is in the mild range [FD minor ,FD moderate ), the fault type is a mild fault; when the fault level is in the moderate range [FD moderate ,FD severe ), the fault type is moderate fault; when the fault level is in the severe range [FD severe ,1], the fault type is severe fault; Construct a fault-tolerant control strategy for each fault type. When the fault type is normal operation, maintain the optimal strategy unchanged. When the fault type is a minor fault, the real-time adsorbent dosage is increased by δ1 based on the optimal strategy; When the fault type is a moderate fault, the real-time adsorbent dosage is increased by δ1 and the real-time exhaust gas flow is reduced by δ2 based on the optimal strategy; when the fault type is a severe fault, the exhaust gas input to the i-th processing unit is stopped and the standby processing unit is activated; According to the fault severity, the fault type of the current i-th processing unit is determined and the corresponding fault-tolerant control strategy is triggered.

7. The method for collecting and treating waste gas from a lithium battery production workshop according to claim 6, characterized in that: The normal range is divided as follows: define the normal operation threshold Among them, η min is the minimum value of processing efficiency, E max ,q max 、F max are the maximum values ​​of treatment energy consumption, adsorbent dosage and waste gas flow rate, respectively, opt 、E opt ,q opt 、F opt are the values ​​of treatment efficiency, treatment energy consumption, adsorbent dosage and exhaust gas flow rate under the optimal operating state; Define Mild Threshold FD minor , and get the normal range [0,FD minor ), the calculation formula of the mild threshold is: FD minor =1.5×FD normal .

8. A system for collecting and treating waste gas from a lithium battery production workshop, which is used to implement the method for collecting and treating waste gas from a lithium battery production workshop as claimed in any one of claims 1 to 7, characterized in that: include: The processing efficiency prediction module is used to collect the historical operation data, adsorbent usage and exhaust gas flow of each processing unit, build a processing efficiency prediction model for each processing unit, and train the processing efficiency prediction model using the historical operation data; Cascade model building module, used to build a cascade model between adjacent treatment units based on historical operation data, adsorbent dosage and exhaust gas flow, form a treatment unit cascade model, obtain the input exhaust gas characteristics of each treatment unit, and define treatment performance indicators; The optimal strategy learning module is used to define the state space, action space and reward function, build the optimization target of each processing unit, learn the optimal strategy of the optimization target by training the Actor-Critic network, and use the optimal strategy as the control strategy of the adaptive feedback controller; A fault degree calculation module is used to monitor the operating status data of each processing unit in real time, predict the real-time processing efficiency of each processing unit using a processing efficiency prediction model, and calculate its fault degree; The fault-tolerant control trigger module is used to classify the fault type into normal operation and abnormal operation according to the fault degree, and to build a fault-tolerant control strategy for each fault type. When the fault type is abnormal operation, the corresponding fault-tolerant control strategy is triggered according to the fault type.

9. An electronic device, characterized in that: It comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the method for collecting and treating waste gas from a lithium battery production workshop as described in any one of claims 1 to 7 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps in the method for collecting and treating waste gas from a lithium battery production workshop as described in any one of claims 1-7.

Images