A method for predicting the hydrophobicity of waterborne polyurethane

Abstract

This invention relates to a method for predicting the hydrophobicity of waterborne polyurethane, comprising the following steps: Step 1, determining a training model for predicting the hydrophobicity of waterborne polyurethane; Step 2, using the training model to predict the hydrophobicity of waterborne polyurethane with an unknown water contact angle; wherein, this invention imports experimental data and physical parameter data of various waterborne polyurethane resins with known water contact angles into the artificial intelligence module of the Matcloud+ materials intelligent big data cloud platform, thereby optimizing the performance standards of the computational simulation prediction system, establishing an artificial intelligence model for predicting the hydrophobicity of waterborne polyurethane with small error and high accuracy, capable of simultaneously predicting the hydrophobicity of multiple waterborne polyurethanes, and also capable of rapidly predicting the hydrophobicity of waterborne polyurethane structures within a larger experimental range, realizing "material design on demand", ultimately reducing experimental costs and shortening the research and development cycle.

Description

Technical Field

[0001] This invention relates to the field of resin hydrophobicity analysis technology, and specifically to a method for predicting the hydrophobicity of waterborne polyurethane. Background Technology

[0002] Waterborne polyurethane (WPU) is a high-molecular polymer produced by the polymerization reaction of polyisocyanates and polyols. It possesses advantages such as being environmentally friendly, having high adhesive strength, good film-forming properties, and adjustable flexibility, and has been widely used in coatings, adhesives, and other fields. Currently, due to the wide variety of waterborne polyurethane resins, significant structural differences, and complex preparation processes, research on the relationship between the molecular structure and preparation process of waterborne polyurethane and its hydrophobicity is mainly limited to experimental analysis. However, this method requires synthesizing the waterborne polyurethane resin from raw materials before testing its hydrophobicity, resulting in high experimental costs, long processing times, heavy workload, and low research efficiency. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for predicting the hydrophobicity of waterborne polyurethane, which can simultaneously predict the hydrophobicity of multiple waterborne polyurethanes, and can also rapidly predict the hydrophobicity of waterborne polyurethane structures within a larger experimental range (including preparation processes).

[0004] The technical solution adopted by this invention to solve its technical problem is: a method for predicting the hydrophobicity of waterborne polyurethane, comprising the following steps:

[0005] Step 1: Determine the training model for predicting the hydrophobicity of waterborne polyurethane;

[0006] 1.1 Import the formulation data of 30 to 1500 waterborne polyurethane resins with known water contact angles into the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform;

[0007] 1.2 In the Artificial Intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, select all data columns except the water contact angle column for the label column, and select the water contact angle data column for the feature column;

[0008] 1.3 In the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, the proportion of test data input is 0.05 to 0.5;

[0009] 1.4. In the artificial intelligence module of the Matcloud+ materials intelligent big data cloud platform, different artificial intelligence algorithms are called for calculation, and the mean absolute difference (MAE), mean square error (MSE), root mean square error (RMSE), and absolute coefficient (R) of different artificial intelligence algorithm models are obtained respectively. 2 );

[0010] 1.5 Select the artificial intelligence algorithm model with the smallest error in step 1.4 as the training model for predicting the unknown water contact angle of waterborne polyurethane.

[0011] Step 2: Use the trained model to predict the hydrophobicity of waterborne polyurethane with unknown water contact angle;

[0012] 2.1 Import the raw material data, preparation process data, and physical parameter data of 1 to 750 types of waterborne polyurethane with unknown water contact angles into the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform.

[0013] 2.2. The training model in step 1.5 is used to calculate the water contact angle of the waterborne polyurethane with an unknown water contact angle.

[0014] Furthermore, the formulation data of the waterborne polyurethane resin mentioned in step 1.1 includes raw material data, preparation process data, water contact angle data, and physical parameter data of the raw materials.

[0015] Furthermore, the raw material data includes the type and quantity of raw materials.

[0016] Furthermore, the preparation process data includes the feeding method of the synthesis reaction, the reaction time of the synthesis reaction, the temperature of the synthesis reaction, the stirring speed of the synthesis reaction, the emulsification time, the emulsification temperature, and the emulsification stirring speed.

[0017] Furthermore, the physical parameter data of the raw materials includes one or more of the molecular orbital data of hydroxyl compounds, molecular orbital data of isocyanate compounds, and molecular orbital data of amine chain extenders.

[0018] Furthermore, the molecular orbital data includes the energy of the lowest empty orbital, the energy of the highest occupied orbital, and the energy difference between the lowest empty orbital and the highest occupied orbital.

[0019] Furthermore, the artificial intelligence algorithms include: decision tree algorithm, support vector regression algorithm, linear regression algorithm, Bayesian regression algorithm, neural network algorithm, and random forest algorithm.

[0020] Furthermore, the smaller the values ​​of the mean absolute difference (MAE), mean square error (MSE), and root mean square error (RMSE), and the smaller the absolute coefficient (R), the better. 2 The closer the value of ) is to 1, the smaller the error.

[0021] The beneficial effects of this invention are:

[0022] This invention discloses a method for predicting the hydrophobicity of waterborne polyurethane. The method involves importing experimental data and physical parameters of various waterborne polyurethane resins with known water contact angles into the artificial intelligence module of the Matcloud+ materials intelligent big data cloud platform. This optimizes the performance standards of the computational simulation prediction system, establishing an AI model for predicting the hydrophobicity of waterborne polyurethane with low error and high accuracy. This model can simultaneously predict the hydrophobicity of multiple waterborne polyurethanes and rapidly predict the hydrophobicity of waterborne polyurethane structures within a wider experimental range. It supports innovative research by experimenters and allows for continuous feedback from experimenters to optimize and improve the accuracy of model predictions, achieving "material-on-demand design," ultimately reducing experimental costs and shortening the R&D cycle. Detailed Implementation

[0023] The present invention will now be described in detail. Therefore, only the components relevant to the present invention are shown.

[0024] In the prior art, the hydrophobicity of waterborne polyurethane can generally be determined by measuring the size of the water contact angle of the coating.

[0025] The present invention provides a method for predicting the hydrophobicity of waterborne polyurethane, comprising the following steps:

[0026] Step 1: Determine the training model for predicting the hydrophobicity of waterborne polyurethane;

[0027] 1.1 Import the formulation data of 30 to 1500 waterborne polyurethane resins with known water contact angle values ​​into the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform;

[0028] The formulation data includes raw material data, preparation process data, water contact angle data, and physical parameter data of the raw materials. The raw material data includes the types and amounts of raw materials.

[0029] Raw materials include hydroxyl compounds, isocyanate compounds, solvents, neutralizing agents, amine chain extenders, and catalysts;

[0030] Hydroxyl compounds include one or more of the following: hydroxyethyl acrylate, hydroxyethyl methacrylate, n-butanol, 1,4-butanediol, 1,6-hexanediol, neopentyl glycol, 1,2-propanediol, 1,3-propanediol, trimethylolpropane, polyether glycol, polycaprolactone diol, polybutylene adipate diol, and polycarbonate diol.

[0031] Isocyanate compounds include one or more of isoflurone diisocyanate, pentamethylene diisocyanate, hexamethylene diisocyanate, 2,4-toluene diisocyanate, 2,6-toluene diisocyanate and 4,4′-dicyclohexylmethane diisocyanate.

[0032] Solvents include one or more of acetone, butanone, water, ethyl acetate, butyl acetate, xylene, toluene, and N,N-dimethylformamide;

[0033] Neutralizing agents include one or more of triethylamine, N,N-dimethylethanolamine, triethanolamine, and sodium hydroxide;

[0034] Amine chain extenders include one or more of ethylenediamine, diethylenetriamine, triethylenetetramine, 1,4-butanediamine, and pentanediamine;

[0035] The catalyst includes one or more of the following: dibutyltin dilaurate, stannous octanoate, bismuth isooctanoate, bismuth laurate, and bismuth neodecanoate;

[0036] The preparation process data includes the feeding method of the synthesis reaction, the reaction time of the synthesis reaction, the temperature of the synthesis reaction, the stirring speed of the synthesis reaction, the emulsification time, the emulsification temperature and the emulsification stirring speed. The feeding method of the synthesis reaction includes one or more of the following: one-step method, stepwise method, and dropwise method.

[0037] The physical parameter data of the raw materials include one or more of the molecular orbital data of hydroxyl compounds, molecular orbital data of isocyanate compounds, and molecular orbital data of amine chain extenders; among which, the molecular orbital data includes the energy of the lowest empty orbital, the energy of the highest occupied orbital, and the energy difference between the lowest empty orbital and the highest occupied orbital;

[0038] 1.2 In the Artificial Intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, select all data columns except the water contact angle column for the label column, and select the water contact angle data column for the feature column;

[0039] 1.3 In the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, the proportion of test data input is 0.05 to 0.5;

[0040] 1.4. In the artificial intelligence module of the Matcloud+ materials intelligent big data cloud platform, different artificial intelligence algorithms are called for calculation, and the mean absolute difference (MAE), mean square error (MSE), root mean square error (RMSE), and absolute coefficient (R) of different artificial intelligence algorithm models are obtained respectively. 2 );

[0041] Artificial intelligence algorithms include: decision tree algorithm, support vector regression algorithm, linear regression algorithm, Bayesian regression algorithm, neural network algorithm, and random forest algorithm;

[0042] 1.5. Select the AI ​​algorithm model with the smallest error from step 1.4 as the training model for predicting the unknown water contact angle of waterborne polyurethane; among which, the smaller the values ​​of mean absolute difference (mae), mean square error (mse), and root mean square error (rmse), and the smaller the absolute coefficient (r... 2 The closer the value of ) is to 1, the smaller the error;

[0043] Step 2: Use the trained model to predict the hydrophobicity of waterborne polyurethane with unknown water contact angle;

[0044] 2.1 Import the raw material data, preparation process data, and physical parameter data of 1 to 750 types of waterborne polyurethane with unknown water contact angles into the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform.

[0045] 2.2. The training model in step 1.5 is used to calculate the water contact angle of the waterborne polyurethane with an unknown water contact angle.

[0046] Example

[0047] Step 1: Determine the training model for predicting the hydrophobicity of waterborne polyurethane;

[0048] 1.1 Import the formulation data of 500 waterborne polyurethane resins with known water contact angle values ​​into the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform;

[0049] Due to the large quantity of ingredients, two examples are provided below.

[0050] Formula 1:

[0051] Raw material data: 1,4-Butanediol 2g, dimethylolpropionic acid 7g, polyether glycol 150g, isoflurane diisocyanate 60g, N,N-dimethylformamide 30g, dibutyltin dilaurate 0.1g, triethylamine 5g, ethylenediamine 6g, water 380g.

[0052] Preparation process data: one-step method, synthesis reaction time 3h, synthesis reaction temperature 85℃, synthesis stirring speed 200rpm / min, emulsification time 2h, emulsification temperature 30℃, emulsification stirring speed 800rpm / min.

[0053] Water contact angle data: 74.126 degrees.

[0054] Physical parameter data: The energy of the lowest empty orbital of isoflurane diisocyanate is -6.5539 eV, the energy of the highest occupied orbital is -0.0130 eV, and the energy difference between the lowest empty orbital and the highest occupied orbital is 6.5409 eV; the energy of the lowest empty orbital of 1,4-butanediol is -5.8868 eV, the energy of the highest occupied orbital is 1.0684 eV, and the energy difference between the lowest empty orbital and the highest occupied orbital is 6.9552 eV.

[0055] Formula 2:

[0056] Raw material data: 1,6-hexanediol 2.5g, dimethylolpropionic acid 7g, polycarbonate diol 150g, isoflurone diisocyanate 60g, N,N-dimethylformamide 30g, dibutyltin dilaurate 0.1g, triethylamine 5g, ethylenediamine 6g, water 380g.

[0057] Preparation process data: dropwise addition method (dropwise addition of isoflurane diisocyanate), synthesis reaction time 4h, synthesis reaction temperature 80℃, synthesis stirring speed 250rpm / min, emulsification time 1h, emulsification temperature 30℃, emulsification stirring speed 1000rpm / min.

[0058] Water contact angle data: 96.234 degrees.

[0059] Physical parameters: The energy of the lowest empty orbital of dimethylolpropionic acid is -6.1862 eV, the energy of the highest occupied orbital is -2.0822 eV, and the energy difference between the lowest empty orbital and the highest occupied orbital is 4.1039 eV.

[0060] The remaining 498 formulas will not be described further.

[0061] 1.2 In the Artificial Intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, select all data columns except the water contact angle column for the label column, and select the water contact angle data column for the feature column;

[0062] 1.3 In the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, the proportion of test data input is 0.2;

[0063] 1.4. In the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, the decision tree algorithm, support vector regression algorithm, linear regression algorithm, Bayesian regression algorithm, neural network algorithm, and random forest algorithm were called for calculation. The calculation results are shown in Table 1:

[0064] Table 1. Calculation results of six artificial intelligence algorithms

[0065]

[0066] As shown in Table 1, the Random Forest algorithm model has the smallest mean absolute difference (MAE), mean squared error (MSE), and root mean squared error (RMSE) among the six algorithm models, and its absolute coefficient (r) is the smallest. 2 The value of ) is closest to 1, so the error of the random forest algorithm model is the smallest. Therefore, the random forest algorithm model is selected as the training model and the hydrophobicity of the waterborne polyurethane with unknown water contact angle is calculated.

[0067] Step 2: Use the trained model to predict the hydrophobicity of waterborne polyurethane with unknown water contact angle;

[0068] 2.1 Import the raw material data, preparation process data, and physical parameter data of the raw materials of three types of waterborne polyurethanes with unknown water contact angles (numbered PU-1, PU-2, and PU-3, respectively) into the artificial intelligence module of the Matcloud+ materials intelligent big data cloud platform.

[0069] PU-1:

[0070] Raw material data: Trimethylolpropane 1.5g, Dimethylolbutyric acid 8g, Polyether glycol 150g, Isoflurone diisocyanate 60g, N,N-dimethylformamide 35g, Dibutyltin dilaurate 0.12g, Triethylamine 5.4g, Ethylenediamine 6g, Water 400g.

[0071] Preparation process data: one-step method, synthesis reaction time 4h, synthesis reaction temperature 75℃, synthesis stirring speed 200rpm / min, emulsification time 1h, emulsification temperature 20℃, emulsification stirring speed 1100rpm / min.

[0072] Physical parameters: The energy of the lowest empty orbital of trimethylolpropane is -5.1733 eV, the energy of the highest occupied orbital is 0.2552 eV, and the energy difference between the lowest empty orbital and the highest occupied orbital is 5.4285 eV.

[0073] PU-2:

[0074] Raw material data: 9g dimethylolpropionic acid, 170g polybutylene adipate diol, 30g isoflurane diisocyanate, 20g hexamethylene diisocyanate, 50g acetone, 0.15g stannous octoate, 6.8g triethylamine, 6g pentamethylenediamine, 380g water.

[0075] Preparation process data: one-step method, synthesis reaction time 6h, synthesis reaction temperature 60℃, synthesis stirring speed 250rpm / min, emulsification temperature 40℃, emulsification time 1h, emulsification stirring speed 1000rpm / min.

[0076] Physical parameters: The energy of the lowest empty orbital of hexamethylene diisocyanate is -6.6313 eV, the energy of the highest occupied orbital is 0.0641 eV, and the energy difference between the lowest empty orbital and the highest occupied orbital is 6.6954 eV.

[0077] PU-3:

[0078] Raw material data: 1,4-Butanediol 2g, dimethylolpropionic acid 7.5g, polybutylene adipate diol 100g, polyether diol 20g, 2,4-toluene diisocyanate 59g, methyl ethyl ketone 50g, bismuth isooctanoate 0.2g, N,N-dimethylethanolamine 5g, ethylenediamine 5g, water 380g.

[0079] Preparation process data: Stepwise method (1,4-butanediol is added in the second step); First step synthesis reaction time 2h, synthesis reaction temperature 80℃, synthesis stirring speed 250rpm / min; Second step synthesis reaction time 2h, synthesis reaction temperature 80℃, synthesis stirring speed 200rpm / min; Emulsification temperature 25℃, emulsification time 1h, emulsification stirring speed 1000rpm / min.

[0080] Physical parameters: The energy of the lowest empty orbital of 2,4-toluene diisocyanate is -5.9732 eV, the energy of the highest occupied orbital is -1.9188 eV, and the energy difference between the lowest empty orbital and the highest occupied orbital is 4.0543 eV.

[0081] 2.2 The hydrophobicity of PU-1, PU-2 and PU-3 was calculated using the random forest algorithm model. The water contact angles of PU-1, PU-2 and PU-3 were 79.398°, 97.287° and 86.78°, respectively.

[0082] Comparative Example

[0083] The water contact angle of the waterborne polyurethane resins PU-1, PU-2, and PU-3 in the examples was tested using experimental methods, as shown in the following steps:

[0084] Step 1: Waterborne polyurethane resins numbered PU-1, PU-2 and PU-3 were synthesized using experimental methods;

[0085] In a four-necked flask equipped with a spherical condenser, a nitrogen inlet tube, and a stirrer, 1.5 g of trimethylolpropane, 8 g of dimethylolbutyric acid, 150 g of polyether glycol, 60 g of isoflurane diisocyanate, 35 g of N,N-dimethylformamide, and 0.12 g of dibutyltin dilaurate were added. The mixture was reacted at 75°C for 4 h with a stirring speed of 200 rpm / min. The temperature was then lowered to 60°C, and 5.4 g of triethylamine was added. The reaction was continued for 0.5 h. Subsequently, 400 g of water was added under stirring at 1100 r / min for emulsification. During emulsification, 6 g of ethylenediamine was added for chain extension. The emulsification temperature was 20°C, and the emulsification time was 1 h to obtain an aqueous polyurethane resin, named PU-1A.

[0086] In a four-necked flask equipped with a spherical condenser, a nitrogen inlet tube, and a stirrer, 9g of dimethylolpropionic acid, 170g of polybutylene adipate diol, 30g of isoflurane diisocyanate, 20g of hexamethylene diisocyanate, 50g of acetone, and 0.15g of stannous octoate were added. The reaction was carried out at 60℃ for 6 hours with a stirring speed of 250 rpm / min. The temperature was then lowered to 60℃, and 6.8g of triethylamine was added. The reaction was continued for 0.5 hours, followed by emulsification with 380g of water under stirring at 1000 r / min. During emulsification, 6g of pentamethylenediamine was added for chain extension. The emulsification temperature was 40℃, and the emulsification was carried out for 1 hour. The acetone was removed by vacuum distillation to obtain an aqueous polyurethane resin, named PU-2A.

[0087] In a four-necked flask equipped with a spherical condenser, a nitrogen inlet tube, and a stirrer, 7.5 g of dimethylolpropionic acid, 100 g of polybutylene adipate diol, 20 g of polyether diol, 59 g of 2,4-toluene diisocyanate, 50 g of butanone, and 0.2 g of bismuth isooctanoate were added. The mixture was reacted at 80 °C for 2 h with a stirring speed of 250 rpm / min. Then, 2 g of 1,4-butanediol was added, and the mixture was reacted at 80 °C for 2 h with a stirring speed of 200 rpm / min. The temperature was then lowered to 60 °C, and 5 g of N,N-dimethylethanolamine was added. The mixture was reacted for 0.5 h, and then 380 g of water was added under stirring at 1000 rpm for emulsification. During emulsification, 5 g of ethylenediamine was added for chain extension. The emulsification temperature was 25 °C, and the emulsification time was 1 h. Butanone was removed by vacuum distillation to obtain an aqueous polyurethane resin, named PU-3A.

[0088] Step 2, Prepare the adhesive film

[0089] Pour PU-1A, PU-2A and PU-3A into a polytetrafluoroethylene mold, place at room temperature for 2 days, then bake at 40℃ for 2 days, and let cool naturally for later use.

[0090] Step 3: Use a water contact angle meter to measure the water contact angles of PU-1A, PU-2A, and PU-3A films respectively;

[0091] The water contact angles of PU-1A, PU-2A, and PU-3A films were measured to be 79.343°, 97.018°, and 85.944°, respectively.

[0092] In summary, the water contact angles of PU-1, PU-2, and PU-3 measured by the waterborne polyurethane hydrophobicity prediction method of this invention are 79.398°, 97.287°, and 86.78°, respectively. The actual water contact angles of PU-1A, PU-2A, and PU-3A films measured by the experimental method in the comparative example are 79.343°, 97.018°, and 85.944°, respectively. Comparison with the actual experimental values ​​shows that the waterborne polyurethane hydrophobicity prediction method of this invention has good prediction accuracy. This demonstrates the rationality of the waterborne polyurethane training model design and the reliability of the theoretical calculations. This method provides a way to predict the hydrophobicity of polyurethane resins. Through a highly efficient artificial intelligence method, it can predict the hydrophobicity of polyurethane resins under various raw material composition systems and different preparation processes, greatly saving experimental time and costs.

[0093] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the scope of the present invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for predicting the hydrophobicity of waterborne polyurethane, comprising the following steps: Step 1: Determine the training model for predicting the hydrophobicity of waterborne polyurethane; 1.1 Import the formulation data of 30 to 1500 waterborne polyurethane resins with known water contact angle values ​​into the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform; 1.2 In the Artificial Intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, select all data columns except the water contact angle column for the label column, and select the water contact angle data column for the feature column; 1.3 In the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, the proportion of test data input is 0.05~0.5; 1.

4. In the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform, the random forest algorithm is called for calculation, and the mean absolute difference (MAE), mean square error (MSE), root mean square error (RMSE), and absolute coefficient (R) of different artificial intelligence algorithm models are obtained respectively. 2 ); 1.5 Select the random forest algorithm model from step 1.4 as the training model for predicting the unknown water contact angle of waterborne polyurethane. Step 2: Use the trained model to predict the hydrophobicity of waterborne polyurethane with unknown water contact angle; 2.1 Import the raw material data, preparation process data, and physical parameter data of 1 to 750 types of waterborne polyurethane with unknown water contact angles into the artificial intelligence module of the Matcloud+ Materials Intelligent Big Data Cloud Platform. 2.

2. The training model in step 1.5 is used to calculate the water contact angle of the waterborne polyurethane with an unknown water contact angle. The formulation data of the waterborne polyurethane resin mentioned in step 1.1 includes raw material data, preparation process data, water contact angle data, and physical parameter data of the raw materials. The raw material data includes the types and quantities of raw materials; The preparation process data includes the feeding method of the synthesis reaction, the reaction time of the synthesis reaction, the temperature of the synthesis reaction, the stirring speed of the synthesis reaction, the emulsification time, the emulsification temperature and the emulsification stirring speed; The physical parameter data of the raw materials include one or more of the molecular orbital data of hydroxyl compounds, molecular orbital data of isocyanate compounds, and molecular orbital data of amine chain extenders.

2. The method for predicting the hydrophobicity of waterborne polyurethane according to claim 1, wherein the molecular orbital data includes the energy of the lowest empty orbital, the energy of the highest occupied orbital, and the energy difference between the lowest empty orbital and the highest occupied orbital.

3. According to the method for predicting the hydrophobicity of waterborne polyurethane as described in claim 1, the smaller the values ​​of the mean absolute difference (mae), root mean square error (mse), and root mean square error (rmse), and the smaller the absolute coefficient (r... 2 The closer the value of ) is to 1, the smaller the error.

Images