A system for predicting production scale up and a prediction method thereof
The system addresses scale up challenges in CAM production by using a theoretical and data-driven approach with machine learning and decision trees to optimize process parameters, achieving efficient and cost-effective scale up from lab to pilot production.
Patent Information
- Application Number
- PCT/CN2025/099003
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2024-06-05
- Filing Date
- 2025-06-04
- Publication Date
- 2025-12-11
AI Technical Summary
The scale up of cathode active materials (CAM) production from laboratory to pilot production faces challenges such as equipment, process, and knowledge gaps due to lack of sufficient time and knowledge gaps, which leads to increased costs and disruptions in production due to time pressure and reliance on expert knowledge.
A system comprising a scale up theoretical unit, a data driven unit, and an assistant decision-making unit, utilizing machine learning algorithms and decision trees, to predict and adjust process parameters for seamless scale up from lab to pilot production.
Enables quick and cost-effective scale up of CAM production by leveraging expert knowledge, reducing material consumption and operational disruptions.
Smart Images

Figure CN2025099003_11122025_PF_FP_ABST
Abstract
Description
A SYSTEM FOR PREDICTING PRODUCTION SCALE UP AND A PREDICTION METHOD THEREOFTECHNICAL FIELD
[0001] This invention relates to a system for predicting production scale up, more specifically production scale up of cathode active materials (CAM) for lithium-ions batteries, as well as a prediction method.BACKGROUND
[0002] The development of new cathode active materials (CAM) normally comprises steps of lab scale trial and pilot trial (or trial production in plant) and production conditions vary from lab scale trial to pilot trial due to scale up of production in many aspects such as equipment, process parameters, know-how from experts etc. Moreover, the scale up of production is under time pressure per request of customers and therefore there is no sufficient time to make deep investigation on experiments design. Even the time is enough, several rounds of tests are needed to achieve a satisfying result, which one side increases the consumption raw materials and therefore raises costs of product development, and in the other side, disrupts the normal operations since the equipment in production lines are required for tests. Furthermore, in actual work, knowledge required is highly replying on experts and know-how might be lost due to personnel movement.
[0003] Therefore, it is still required to provide a method and system for scale up of production from lab scale trial to pilot trial that could be done quickly, with low costs and taking full usage of expert knowledges.SUMMARY OF THE INVENTION
[0004] In one aspect, the present invention provides a system for predicting production scale up, comprising units of
[0005] (a) a scale up theoretical unit that predicts process parameters of production scale up with input of experimental data obtained from lab scale trial;
[0006] (b) a data driven unit that adjusts the process parameters obtained from the scale up theoretical unit; and
[0007] (c) an assistant decision-making unit that outputs parameters applicable for trial production in plant for next round of prediction based on the data obtained from data driven unit, wherein said assistant decision-making unit is configured to map knowledge data from one or more knowledge databases including one or more decision trees.
[0008] In another aspect, the present invention provides a method for predicting production scale up, comprising steps of
[0009] (i) input initial value of experimental data from lab scale trial into scale up theoretical unit to obtain predicted process parameters for production scale up,
[0010] (ii) adjust the predicted scale up parameters with data driven unit based on one or more algorithm of machine learning,
[0011] (iii) solve revision direction and amplitude of data obtained from step (ii) through assistant decision-making unit,
[0012] (iv) revised parameters obtained from step (iii) that are applicable for trial production in plant are used for next round of prediction through steps (i) to (iv) , and
[0013] (v) repeat iteration process from steps (i) to (iv) until final predicted results meet with those obtained from reality.
[0014] Surprisingly, by using the invented system and method, the scale up of CAM production could be achieved quickly, with low costs and taking full usage of expert knowledges.DETAILED DESCRIPTION OF THE INVENTION
[0015] The undefined article “a” , “an” , “the” means one or more of the species designated by the term following said article.
[0016] In the context of the present disclosure, any specific values mentioned for a feature (comprising the specific values mentioned in a range as the end point) can be recombined to form a new range.
[0017] In the context of the present disclosure, each aspect so defined may be combined with any other aspect or aspects unless clearly indicated to the contrary. In particular, any feature indicated as being preferred or advantageous may be combined with any other feature or features indicated as being preferred or advantageous.
[0018] In the context of the present disclosure, “process data” means any and all parameter in process of production that could be shown in whatever kind of equipment or instrument and any and all components as raw materials that are mixed for chemical reactions to obtain final products, as well as weight percentage of each component.
[0019] In the context of the present disclosure, “equipment data” means any and all parameter of equipment that is used for producing the final products with addition of raw materials.
[0020] In the context of the present disclosure, “product quality data” means any and all performance tests results on final products of CAM.
[0021] In the context of the present disclosure, “knowledge database” means database of know-how and all kinds of unpublished knowledge obtained from experts in this technical field including one or more decision trees.
[0022] In the context of the present disclosure, “knowledge data” means any and all kinds of data included in knowledge database.
[0023] In the context of the present disclosure, “decision trees” means any and all paths to decide an output parameter based on an input parameter and both input and output parameters are knowledge data.
[0024] The present invention provides a system for predicting production scale up, comprising units of
[0025] (a) a scale up theoretical unit that predicts process parameters for production scale up with input of experimental data obtained from lab scale trial;
[0026] (b) a data driven unit that adjusts the process parameters obtained from the scale up theoretical unit; and
[0027] (c) an assistant decision-making unit that outputs parameters applicable for trial production in plant for next round of prediction based on the data obtained from data driven unit, wherein said assistant decision-making unit is configured to map knowledge data from one or more knowledge databases including one or more decision trees.
[0028] The present invention also provides a method for predicting production scale up, comprising steps of
[0029] (i) input initial value of experimental data from lab scale trial into scale up theoretical unit to obtain predicted process parameters for production scale up,
[0030] (ii) adjust the predicted scale up parameters with data driven unit based on one or more algorithm of machine learning,
[0031] (iii) solve revision direction and amplitude of data obtained from step (ii) through assistant decision-making unit,
[0032] (iv) revised parameters obtained from step (iii) that are applicable for trial production in plant are used for next round of prediction through steps (i) to (iv) , and
[0033] (v) repeat iteration process from steps (i) to (iv) until final predicted results meet with those obtained from reality.
[0034] The scale up theoretical units are running based various scale up models such as constant Reynolds numbers (Re) , constant energy inputs per unit volume (P / V) , constant linear rate and constant Froude number.
[0035] Reynolds number (Re) in fluid dynamics is a dimensionless quantity that helps predict fluid flow patterns in different situations by measuring the ratio between inertial and viscous forces. Re is define as Re = uL / ν = ρuL / μ, where ρ is the density of the fluid, u is the flow speed, L is the characteristic length, μ is the dynamic viscosity of the fluid, v is the kinematic viscosity of the fluid. In order to maintain the similarity between the two flow systems under the action of viscous force, the ratio of the inertial force of the water flow to the viscous force must be equal, that is, the Reynolds number must be equal.
[0036] At constant tip speed, the impeller Reynolds number increases as equipment size becomes larger. Keeping Reynolds number constant for scale up of either pipe flow or impeller mixing will result in lower velocities in the larger diameter pipe or vessel. This reduces flow and the effectiveness of either the pipe or mixer. And other parameters need to be adjusted to offset the brought drawbacks.
[0037] Froude number is a dimensionless number in continuum mechanics defined as the ratio of the flow inertia to the external field. The Froude number is based on the speed length ratio defined as Fr = u / (gL) 1 / 2, where u is the local flow velocity, g is the local external field and L is a characteristic length.
[0038] For different equipment and sections, any one of above-list scale up models could be used or more than one scale up models could be combined to use. In some preferred embodiments, scale up models of constant Re, constant P / V, constant tip speed and constant Froude Number are used simultaneously.
[0039] The data input to scale up is from lab scale trial including process data, equipment data and product quality data. In some embodiments, said equipment data comprises type of equipment, size of equipment such as diameter, height and volume etc., and inner part of equipment such as agitator and baffle etc. Said process data comprises raw materials, temperature, pressure, stirring speed and mixing time etc. Said product quality data comprises particle size distribution (PSD) , residue Li (%) , pH value, capacity, C-rate and cycle performance etc.
[0040] The predicted scale up process parameters obtained from scale up theoretical unit could be input into data driven unit that helps to further correct coefficients. The data driven unit is based on machine learning that takes use of at least one algorithm selected from a group consisting of Boost algorithm, SVM (Support Vector Machine) algorithm, neuron network algorithm, Bayes algorithm, decision tree algorithm, LSH (Locality Sensitive Hashing) algorithm and KNN (K-Nearest Neighbor) algorithm. With the output data from the scale up theoretical unit and combined with information of raw materials and historical data from both labs and product lines, the closest prediction values could be found through similarity and distance measurement and the data obtained from data driven unit are ready for input into assistant decision-making unit.
[0041] In some preferred embodiments, both scale up theoretical unit and data driven unit are used before assistant decision-making unit in the sequence of scale up theoretical unit first and following data driven unit.
[0042] The assistant decision-making unit is established based on knowledge database including one or more decision trees that are built up by interviewing experts in this technical area and having a structure according to flowchart of process for preparing cathode active materials used in batteries. After the data obtained from data driven unit is input into assistant decision-making unit, combined with historical data from production lines, the direction and amplitude of parameters to be adjusted could be solved by calculations along with branches of the tree to obtain the optimal solutions. In some embodiments, the output data from scale up theoretical unit could be directly input into assistant decision-making unit.
[0043] The output data from assistant decision-making unit could be used for next round of prediction. After several rounds of iteration calculation, the obtained data is much closer to the actual situations and the iteration will be continued until the errors of parameters are falling in acceptable ranges. Whether final predicted results meet with actual parameters is determined by Coefficient of Determination R2 according to Formula I: R2 = 1 –Σi (yi-yf) 2 / Σi (yi-ya) 2 Formula 1
[0044] wherein, yi is a data point in reality, yf is the data point in prediction corresponding to yi and ya is an average value of yi and when R2 is in a range of from 0.8 to 1 and preferably from 0.9 to 1, the predicted results are acceptable.
[0045] In some preferred embodiments, the invented method for predicting scale up of production comprises steps of
[0046] (i) input initial value of data from lab scale trial into scale up theoretical unit to obtain predicted scale up parameters,
[0047] (ii) adjust the predicted scale up parameters with data driven unit based on one or more algorithm of machine learning,
[0048] (iii) solve revision direction and amplitude of data obtained from step (ii) through assistant decision-making unit,
[0049] (iv) revised parameters obtained from step (iii) that are applicable for trial production in plant are used for next round of prediction through steps (i) to (iv) , and
[0050] (v) repeat iteration process from steps (i) to (iv) until final predicted results meet with those obtained from reality.
[0051] By using the invented prediction method for production scale up based on the system for predicting production scale up, the scale up of CAM production could be achieved quickly, with low costs and taking full usage of expert knowledges.
[0052] BRIEF DESCRIPTION OF FIGURES
[0053] FIG. 1 shows a schematic diagram of the scale up of production from lab scale trial to trial production in plant based on the invented system for predicting production scale up by using the invented prediction method.
[0054] FIG. 2 shows a schematic diagram that illustrates how “Scale up Theoretical Unit” works.
[0055] FIG. 3 shows a schematic diagram that illustrates how “Data Driven Unit” works.
[0056] FIG. 4 shows a schematic diagram that illustrates how “Assistant Decision-making Unit” works.
[0057] EMBODIMENT
[0058] The present invention may be embodied in many different forms, one or more specific embodiments being illustrated in the drawings and the following description. It is to be understood that such disclosure is illustrative of the principles of the invention and is not intended to limit the invention to the specific embodiments shown and described.
[0059] Embodiment 1
[0060] A system for predicting production scale up, comprising units of
[0061] (a) a scale up theoretical unit that predicts process parameters for production scale up with input of experimental data obtained from lab scale trial;
[0062] (b) a data driven unit that adjusts the process parameters obtained from the scale up theoretical unit; and
[0063] (c) an assistant decision-making unit that outputs parameters applicable for trial production in plant for next round of prediction based on the data obtained from scale up theoretical unit or data driven unit, wherein said assistant decision-making unit is configured to map knowledge data from one or more knowledge databases including one or more decision trees.
[0064] Embodiment 2
[0065] The system according to Embodiment 1, wherein said scale up theoretical unit provides predictions based on scale up theories that follows at least one principle selected from a group consisting of constant Reynolds numbers, constant energy inputs per unit volume, constant linear rate and constant Froude number.
[0066] Embodiment 3
[0067] The system according to any one of Embodiments 1 to 2, wherein said data driven unit is based on machine learning that takes use of at least one algorithm selected from a group consisting of Boost algorithm, SVM (Support Vector Machine) algorithm, neuron network algorithm, Bayes algorithm, decision tree algorithm, LSH (Locality Sensitive Hashing) algorithm and KNN (K-Nearest Neighbor) algorithm.
[0068] Embodiment 4
[0069] An application of the system for predicting production scale up according to any one of Embodiments 1 to 3 in predicting scale up of cathode active materials (CAM) production in lithium-ions batteries.
[0070] Embodiment 5
[0071] The application according to Embodiment 4, wherein the input and output data comprises parameters set in equipment for each step of preparing CAM, formulations of raw materials, quality analysis results of prepared CAM samples or products.
[0072] Embodiment 6
[0073] A method for predicting scale up of production, comprising steps of
[0074] (i) input initial value of data from lab scale trial into scale up theoretical unit to obtain predicted scale up parameters,
[0075] (ii) adjust the predicted scale up parameters with data driven unit based on one or more algorithm of machine learning,
[0076] (iii) solve revision direction and amplitude of data obtained from step (ii) through assistant decision-making unit,
[0077] (iv) revised parameters obtained from step (iii) that are applicable for trial production in plant are used for next round of prediction through steps (i) to (iv) , and
[0078] (v) repeat iteration process from steps (i) to (iv) until final predicted results meet with those obtained from reality.
[0079] Embodiment 7
[0080] The method for predicting scale up of production according to Embodiment 6, wherein said scale up theoretical unit makes predictions based on scale up mechanism unit that follows at least one principle selected from a group consisting of constant Reynolds numbers, constant energy inputs per unit volume, constant linear rate and constant Froude number.
[0081] Embodiment 8
[0082] The method for predicting scale up of production according to any one of Embodiments 6 to 7, wherein said data driven unit is based on machine learning that takes use of at least one algorithm selected from a group consisting of Boost algorithm, SVM (Support Vector Machine) algorithm, neuron network algorithm, Bayes algorithm, decision tree algorithm, LSH (Locality Sensitive Hashing) algorithm and KNN (K-Nearest Neighbor) algorithm.
[0083] Embodiment 9
[0084] The method for predicting scale up of production according to any one of Embodiments 6 to 8, wherein said assistant decision-making unit is configured to map knowledge data from one or more knowledge databases including one or more decision trees.
[0085] Embodiment 10
[0086] The method for predicting scale up of production according to any one of Embodiments 6 to 9, wherein in step (v) , whether final predicted results meet with actual parameters is determined by Coefficient of Determination R2 according to Formula I: R2 = 1 –Σi (yi-yf) 2 / Σi (yi-ya) 2 Formula 1
[0087] wherein, yi is a data point in reality, yf is the data point in prediction corresponding to yi and ya is an average value of yi and when R2 is in a range of from 0.8 to 1 and preferably from 0.9 to 1, the predicted results are acceptable.
[0088] Embodiment 11
[0089] An application of the method for predicting scale up of production according to any one of Embodiments 6 to 10 in predicting scale up of CAM (cathode active materials) production in lithium-ions batteries.
[0090] Embodiment 12
[0091] The application according to Embodiment 11, wherein the input and output data comprises parameters set in equipment for each step of preparing CAM, formulations of raw materials, quality analysis results of prepared CAM samples or products.
[0092] The batch process data, batch equipment data and batch product quality data obtained from lab scale trial is input into “Scale up Theoretical Unit” to generate predicted scale up parameters that is to be input into “Data Driven Unit” . The adjusted parameters obtained from “Data Driven Unit” is input into “Assistant Decision-making Unit” to generate advised control parameters that could be used in trial production in plant. And the batch process data, batch equipment data and batch product quality data obtained from the trial production in plant could be used to improve the parameters setting in lab scale trial. Such process could be iterated to continuously optimize the system for production scale up.
[0093] The information of equipment, process parameters and product quality obtained from lab scale trial turn to that could be used in trial production in plant through scale up theoretical unit including but not limited to constant Reynolds numbers (Re) , constant energy inputs per unit volume (P / V) , constant linear rate and constant Froude number. The subscript “B” means “B sample” (i.e. the sample obtained from lab scale trial) and the subscript “C” means “C sample” (i.e. the sample obtained from trial production in plant) .
[0094] Parameters obtained from scale up theoretical unit, quality requirements for pilot products, parameters of samples from lab as well as quality test results of samples from lab are input to “Data Driven Unit” to obtain predicted process parameters for trial production in plant. A database of existing products including information of raw materials, process parameters and product qualities is used for “Data Driven Unit” combined with one or more algorithms such as Boost algorithm, SVM (Support Vector Machine) algorithm, neuron network algorithm, Bayes algorithm, decision tree algorithm, LSH (Locality Sensitive Hashing) algorithm and KNN (K-Nearest Neighbor) algorithm. Said database of existing products is updated lively to improve the accuracy of simulation.
[0095] The key issue of “Assistant Decision-making Unit” is knowledge database including one or more decision trees. The target parameters are product qualities and once a product quality parameter is focused, the knowledge tree provides all correlated process parameters including positively and negatively correlated ones in different degrees. And after the critical process parameter is found, one or more control parameter (s) are identified to be used in trial production in plant. Here the term of “product quality parameter” means the goal of performance of products to be achieved, the term of “process parameter” means any factor (s) influencing the process of preparing the products and the term of “control parameter” means any parameter that is controllable or readable from instruments used in preparation of products. The process parameter could be realized by one or more control parameter (s) and sometimes the process parameter is equal to control parameter e.g. temperature or time.
[0096] In production of cathode active materials (CAM) in lithium-ions batteries, it is required to quickly adjust and continuously optimize process parameters in water washing process of a pilot production line due to different products specifications and the gap between equipment used in labs and pilot production lines. Equipment for a typically washing process include (i) . a metering bin for semi-finished products of cathode active materials with a loss-in-weight metering unit to measure the weight of the materials; (ii) . a pure water storage tank with a cooling coil to provide cooling water to the washing kettle; (iii) . an online liquid flowmeter between the pure water storage tank and the washing kettle to check the mass flow rate of the pure water; and (iv) . a washing kettle connected to the metering bin and the pure water storage tank and equipped with a temperature detecting device to check the temperature of water washing and a stirring device to control the stirring paddle speed.
[0097] EXAMPLE
[0098] The following describes an example of the system for predicting production scale up according to the present invention applied to a pilot production line of water washing process for CAM (cathode active materials) in lithium-ions batteries. Various modifications and variations conceivable by those skilled in the art can be made without departing from the scope or spirit of the present disclosure. This example shall be considered illustrative only, and the protection scope of the present invention is to be specified by the appended claims and their equivalents.
[0099] The control parameters in this example are temperature and weight of pure water, weight of the water washing material, weight of the rinsing water, and stirring frequency. Here the control parameters are equal to process parameters. The product quality parameters comprise quality parameters of the material after the washing process, such as moisture, specific surface area, residual alkali and magnetic impurities.
[0100] It is required to classify scale up theoretical units according to different process steps. The different operating parameters in each process step influence the parameters in mechanisms and product properties thereof. Applicable formula is determined according to the mechanism of a specific unit and the process parameters of B sample (i.e. the sample obtained from lab scale trial) are input, thus the corresponding operating parameters of C sample (i.e. the sample obtained from trial production in plant) are calculated through equivalent amplification according to the principle of invariance of the mechanism parameters.
[0101] In the washing process, the equipment used in lab is different from those used in trial production in plant, the mechanism of constant Froude number is used to ensure that the mixing conditions stay unchanged from equipment in lab to that in trial production in plant. Given the stirring speed, the stirring paddle diameter of B sample and the stirring paddle diameter of C sample, the theoretical value of stirring speed of C sample can be calculated through the mechanistic of constant Froude number.
[0102] To train the “Data Driven Unit” , it is required to collect data according to various product types including the actual values of all process parameters and quality test results of B and C samples and the theoretical values of process parameters of C samples obtained from “Scale up Theoretical Unit” (collectively “historical data” ) . All historical data is transformed into a readable format for machine learning algorithms such as LSH (Locality Sensitive Hash) . The target parameter is actual stirring speed in trial production in plant. With the iteration of algorithms and adjustment of algorithm parameters, the predicted value is converging to the actual value of stirring speed i.e. the coefficient of determination of R2 should be as close to 1 as possible.
[0103] With a requirement of actual product scale up, the system for predicting production scale up is utilized to predict the mixing frequency by steps of:
[0104] (1) Calculating the theoretical stirring speed of C sample by scale up theoretical model based on the actual stirring speed of B sample;
[0105] (2) Inputting the actual stirring speed of B sample, the actual test results of product quality of B sample, the theoretical stirring speed of C sample and the target product quality parameters of C sample into Data Driven Unit in a readable format for machine learning algorithms such as LSH and outputting an optimized theoretical stirring speed of C sample;
[0106] (3) Applying the optimized theoretical stirring speed to actual trial production in plant to verify the reliability; and
[0107] (4) Repeating above steps until the error in test results of product quality is within 5%deviation.
[0108] When a new product is generated, the database of existing products for training Data Driven Unit must be updated timely to ensure the accuracy of the data source.
[0109] The assistant decision-making unit is based on knowledge mapping by interviewing experts. The product quality parameters, process parameters, and control parameters are collected according to the production process of CAM and the correlations among those parameters are explored to form knowledge database and further a complete decision-making assisted system.
[0110] After the stirring speed obtained from Data Driven Unit is verified in the trial production in plant, it is needed to identify the deviations e.g. high residual alkali, and input such deviation into Assistant Decision-making Unit for searching. Based on the process parameters related to the residual alkali index, Assistant Decision-making Unit searches for the process parameters to be adjusted on production lines along the branches of the knowledge tree and gives adjustment direction for such process parameters. The adjustment recommendations include decreasing stirring speed, reducing the weight of intermediates, increasing the mixing time, etc. And the operator may decide which recommendation(s) could be taken according to actual conditions of production lines.
Claims
1.A system for predicting production scale up, comprising units of(a) a scale up theoretical unit that predicts process parameters for production scale up with input of experimental data obtained from lab scale trial;(b) a data driven unit that adjusts the process parameters obtained from the scale up theoretical unit; and(c) an assistant decision-making unit that outputs parameters applicable for trial production in plant for next round of prediction based on the data obtained from data driven unit,wherein said assistant decision-making unit is configured to map knowledge data from one or more knowledge databases including one or more decision trees.2.The system according to Claim 1, wherein said scale up theoretical unit provides predictions based on at least one principle selected from a group consisting of constant Reynolds numbers, constant energy inputs per unit volume, constant linear rate and constant Froude number.3.The system according to any one of Claims 1 to 2, wherein said data driven unit is based on machine learning that takes use of at least one algorithm selected from a group consisting of Boost algorithm, SVM (Support Vector Machine) algorithm, neuron network algorithm, Bayes algorithm, decision tree algorithm, LSH (Locality Sensitive Hashing) algorithm and KNN (K-Nearest Neighbor) algorithm.4.An application of the system for predicting production scale up according to any one of Claims 1 to 3 in predicting scale up of cathode active materials (CAM) production in lithium-ions batteries.5.The application according to Claim 4, wherein the input and output data comprises parameters set in equipment for each step of preparing CAM, formulations of raw materials, quality analysis results of prepared CAM samples or products.6.A method for predicting scale up of production, comprising steps of(i) input initial value of data from lab scale trial into scale up theoretical unit to obtain predicted scale up parameters,(ii) adjust the predicted scale up parameters with data driven unit based on one or more algorithm of machine learning,(iii) solve revision direction and amplitude of data obtained from step (ii) through assistant decision-making unit,(iv) revised parameters obtained from step (iii) that are applicable for trial production in plant are used for next round of prediction through steps (i) to (iv) , and(v) repeat iteration process from steps (i) to (iv) until final predicted results meet with those obtained from reality.7.The method for predicting scale up of production according to Claim 6, wherein said scale up theoretical unit makes predictions based on at least one principle selected from a group consisting of constant Reynolds numbers, constant energy inputs per unit volume, constant linear rate and constant Froude number.8.The method for predicting scale up of production according to any one of Claims 6 to 7, wherein said data driven unit is based on machine learning that takes use of at least one algorithm selected from a group consisting of Boost algorithm, SVM (Support Vector Machine) algorithm, neuron network algorithm, Bayes algorithm, decision tree algorithm, LSH (Locality Sensitive Hashing) algorithm and KNN (K-Nearest Neighbor) algorithm.9.The method for predicting scale up of production according to any one of Claims 6 to 8, wherein said assistant decision-making unit is configured to map knowledge data from one or more knowledge databases including one or more decision trees.10.The method for predicting scale up of production according to any one of Claims 6 to 9, wherein in step (v) , whether final predicted results meet with actual parameters is determined by Coefficient of Determination R2 according to Formula I: R2 = 1 –Σi (yi-yf) 2 / Σi (yi-ya) 2 Formula 1wherein, yi is a data point in reality, yf is the data point in prediction corresponding to yi and ya is an average value of yi and when R2 is in a range of from 0.8 to 1 and preferably from 0.9 to 1, the predicted results are acceptable.11.An application of the method for predicting production scale up according to any one of Claims 6 to 10 in predicting scale up of CAM (cathode active materials) production in lithium-ions batteries.12.The application according to Claim 11, wherein the input and output data comprises parameters set in equipment for each step of preparing CAM, formulations of raw materials, quality analysis results of prepared CAM samples or products.
Citation Information
Patent Citations
Battery production workflow optimization
US20230261188A1
Cited By
Micro centrifugal pump blade setting angle correction method based on Reynolds number sensitivity analysis
CN122154562A