Method for a continuous, automated and integrated prediction of a first output and a second output in a palm oil mill

Abstract

The present invention provides a method for a continuous, automated and integrated prediction of a first output and a second output in real-time in a palm oil mill, whereby the method is a predictive model or equation between a first input parameter, a second input parameter, a third input parameter and a fourth input parameter.

Description

[0001] METHOD FOR A CONTINUOUS, AUTOMATED AND INTEGRATED PREDICTION OF A FIRST OUTPUT AND A SECOND OUTPUT IN A PALM OIL MILL FIELD OF INVENTION The present invention generally relates to a real-time prediction method. Specifically, this present invention relates to a method of continuous, automated and integrated prediction of a first output and a second output in real-time. This approach involves an intricate predictive model or equation achieved through a machine learning algorithm, which incorporates various combinations and interactions of four input parameters: a first input parameter, a second input parameter, a third input parameter, and a fourth input parameter. BACKGROUND OF INVENTION In a typical palm oil milling process, numerous steps are undertaken to obtain palm oil from fresh fruit bunches (FFB). These steps include sterilization, stripping, digestion and pressing, clarification, purification, drying, and storage. [Source: Flow chart for the palm oil process - https: / / lipidlibrary.aocs.org / edible-oil-processing / palm-oil] Freshly harvested fruit bunches and individual loose fruits are transported to the palm oil mill for further processing. The main objective at this stage is to sterilize the palm fruitlets in order to prevent the activity of lipolytic enzymes. These enzymes can degrade the quality of the oil by increasing the levels of free fatty acids. This degradation occurs as a result of the bruising and damage caused to the fruitlets during the harvesting and transportation process to the mill. Sterilization serves several purposes. Firstly, it helps to separate the individual fruits from the bunches, making them easier to handle during subsequent steps. Additionally, sterilization softens the fruits, making it easier to extract oil from them. It also conditions the kernels and prompts the coagulation of proteins, further facilitating the oil extraction process. The sterilization process involves the use of live steam, which is delivered at a pressure of approximately 2.75 to 3.1 bars for a specific duration of time. In a conventional palm oil mill, the bunches and loose fruits are loaded into cages and then placed in sterilizers. These sterilizers can accommodate batches of bunches at a time. The cooking of the fruits takes place by introducing steam into the sterilizer through inlet valves. This steam is regulated and controlled through an automated program, ensuring consistent and efficient sterilization. Sterilization plays a crucial role in palm oil mills as a vital process. Over time, there have been advancements and adaptations in sterilization methods. The two most widely employed techniques include the continuous sterilizer and the vertical sterilizer. The process of removing (stripping) the fruitlets from the sterilized bunches (threshing) is accomplished in a rotating enclosure with bars, which permit the fruits to pass through while retaining the empty clusters. The dislodged fruits are gathered by a conveyor located beneath the enclosure, while the empty bunches emerge at the end of the enclosure. The extracted fruitlets are subsequently introduced into a well-insulated cylindrical tank, known as a digester, which is maintained at a temperature ranging from 85°C to 95°C through the injection of live steam. This digester is equipped with beater arms that break down the fruit and release the oil. The digested mixture of fruit, palm nuts, and residue is then continuously supplied to a screw press. This press yields a liquid flow consisting of oil, fine particles, and an aqueous phase, as well as a solid cake comprising the fibrous remains of the fruit and the kernels of the palm nuts. The liquid flow is directed to a settling tank via a vibrating screen, which returns any retained particles back to the digester. The oil collected from the settling tank undergoes purification in a purifier to further eliminate impurities, followed by drying in a vacuum dryer. The sludge that accumulates in the settling tank is directed to a decanter, where it is separated into a dense or heavy effluent phase and a lighter, oily phase that can be recovered for further use. The palm oil screw press comprises of a tubular structure with holes and a closely fitting pair of rotating screws that move in opposite directions within it. These screw presses are highly regarded for their ability to effectively rupture the unopened oil cells within the fruit pulp through the agitation and squeezing action that takes place in the pressing enclosure, resulting in higher oil extraction. They can be considered as an additional step to the digestion process. They are particularly suitable for extracting palm oil and various other types of vegetable oil. The pressure within the press is typically regulated automatically by a hydraulic system. The pressure level of the cone determines the amount of force applied to the mesh. The hydraulic system receives feedback from the power consumption of the screw press motor (measured in amperes). The appropriate pressure is achieved when the nut breakage is less than 20% and the oil content in the press fibre is approximately 7.5%. Setting the pressure too high leads to increased nut breakage and loss of kernel. Once the pressing is complete, the oil palm fruit is separated into two components: crude palm oil, which is a mixture of oil, water, and solid impurities, and the press cake, which consists of fibre and nut. The extracted oil contains fruit fragments, water, sand, and dirt. The press cake consists of palm fibre and nuts. The remaining fibres and nuts are retrieved and dried. [Source: https: / / www.viridis-engineering.com / solutions / palm-oil / palm-oil-mills / digester-and-pressing / ] The process of digestion greatly influences the effectiveness of the pressing stage, as an ineffective digestion will result in the oil remaining trapped in the cells and being unable to be extracted during pressing. According to industry standards, a digester should be filled and maintained at a level of at least three quarters of its total volume (preferably more than ¾ or 75%) in order to achieve optimal performance. The volume of the main palm digester (MPD) within the digester directly affects the efficiency of the stirring effect. It has been observed that when the digester is filled to ¾ (75%) capacity with MPD, the stirring effect operates at 55% efficiency, but this efficiency decreases to 35% when the digester is only filled to 60% capacity with MPD. Consequently, the inefficiency in stirring directly results in increased losses of oil. Maintaining the digester at the optimum level allows for a reduction in oil loss. Additionally, the main screw speed and cone pressure play a significant role in achieving a successful pressing mechanism. One of the most important conventions when operating a screw press is to be attentive to the discharge from the press and make necessary adjustments. To maintain oil loss at an acceptable level, the optimal hydraulic pressure should not be set above 70 bar. This range is based on industry standards and can be found in screw press models with motor loads between 25 amperes and 50 amperes (P10 screw press), 45 amperes and 50 amperes (P15 screw press) and 60 amperes and 65 amperes (P20 screw press). Additionally, regular maintenance is important to keep the screw press in good working order. [Source: Screw Press Operation Optimisation for Oil and Kernel Recovery Enhancement, Palm Oil Engineering Bulletin No.130 – Feature Article] The current OLWB in a pressed mesocarp fibre based on Malaysian Palm Oil Board (MPOB) data is within a range of between 4.00% to 6.20%. Oil extraction efficiency at the pressing station is determined by measuring how much oil remains in the pressed mesocarp fibre of the press cake. On a dry basis, this would be 5% to 11%. On a wet basis, this would be 1.8% to 3.96%. [Source: Journal of Oil Palm Research Vol.25 (1) April 2013 p.116-122]

[0002] SUMMARY OF INVENTION The present invention generally relates to a real-time prediction method. This present invention provides a method to continuously, automatically and seamlessly predict a first output and a second output in real-time within a palm oil mill, whereby the method is a predictive model or equation which connects a set of input parameters, including a first input parameter, a second input parameter, a third input parameter and a fourth input parameter. More specifically, the present invention pertains to a proactive approach for predicting the oil loss in a pressed mesocarp fibre and broken nuts found within a pressed cake in a palm oil mill. The prediction process seamlessly integrates with the operational workflow, providing instant analysis and insights.

[0003] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS OF THE PRESENT INVENTION The present invention generally relates to a real-time prediction method. It can be appreciated that the parameters for the present invention are not obvious for a person skilled in the art. A person skilled in the field would recognize that the parameters for this invention are not readily obvious and have been carefully determined by the inventors through extensive trials, observations, discussions, and collective expertise. It would have required considerable effort and analysis to arrive at these parameters. None of the prior arts mentioned specifically describe the method of this present invention. The oil palm fruitlets move to the pressing station for pressing after a series of processes in the oil palm mills (such as sterilizing, threshing, and digestion). The oil palm fruits are separated into two categories after being pressed in the pressing station: (a) a mixture of oil, water, and non-solid (NOS) impurities; and (b) press cake (consisting of the pressed mesocarp fibre and nuts or kernels). The correct pressing pressure is applied when a small percentage of broken nuts is obtained, preferably <20% per industry standards and the oil loss on a wet basis (OLWB) of the pressed mesocarp fibre is <4.5%. If the pressure is set too high the nut breakage increases and kernel is lost. If the pressure is set too low, the remaining oil in the mesocarp fibre will be high and oil is lost. According to industry standards, the right pressing pressure is used when there is only a limited amount of cracked nuts, preferably less than 20%, and the oil loss on a wet basis (OLWB) of the pressed mesocarp fiber is less than 4.5%. If the pressure is adjusted too high, the likelihood of nut breakage and kernel loss increases. Whereas, the amount of leftover oil in the mesocarp fibre will be high and oil is lost if the pressure is adjusted too low. For the purposes of this invention, "OLWB" refers to the estimation or measurement of oil losses in the pressed mesocarp fibre during the solvent extraction process of palm oil, where the oil content is expressed as a percentage of wet weight or total weight of the pressed fibre. Currently, achieving low OLWB, or having little oil left in the press mesocarp fibre after pressing, determines how effectively oil is extracted at the press station. The traditional approach to measuring OLWB is by collecting samples every two hours and analyzing them using a portable near-infrared (NIR) spectrometer. This method is time-consuming, prone to errors, and not even appropriate for up- scaling or optimization. In order to create a real-time prediction system, data on a number of variables, including cone press pressure, screw press speed, the level of the digester in the palm oil mill, and the opening or closing of the bottom drainage, are collected. The data from the mill is used to train a machine learning algorithm, which is then used to create the said prediction model, which can predict the OLWB and broken nuts in a pressed cake in real-time. The following were noted based on the knowledge, observation, and testing by the inventors to arrive at the method of the present invention as efficiently based on expected results:- The purpose of the present invention is to provide a method for continuous, automated, and integrated prediction of a first output (percentage of OLWB in a pressed mesocarp fibre) and a second output (percentage of broken nuts in a pressed cake of a pressed mesocarp fibre) in real-time to allow modifications to be made in the palm oil mill to guarantee that the broken nuts are less than 20% and the OLWB is less than 4.5%, respectively, in accordance with current industry standards. A machine learning algorithm that is easily trained using fresh samples (oil palm fresh fruit bunches / fruitlets collected from the estates, as well as the process conditions in the palm oil mill, is used to customize and generate the prediction model or equation of the present invention. In order to lessen reliance on humans for monitoring purposes, the present invention is advancing toward automation in the palm oil milling process. Human error is eliminated by the current innovation because an operator is not required to manually monitor and adjust variables like the digester level and cone press speed. The Supervisory Control and Data Acquisition (SCADA) system will monitor, regulate, correct, and store each of these characteristics. With the present invention, it is possible to do away with the harmful chemicals which are dangerous to human health like n-hexane that are typically used to measure OLWB and broken nut percentages. Objectives of the present invention: This present invention relates generally to a method of continuous, automated and integrated prediction of a first output and a second output in real-time, whereby the method is a predictive model or equation between a combination of a first input parameter, a second input parameter, a third input parameter and a fourth input parameter. More specifically, the present invention relates to a method of continuous, automated and integrated prediction of oil loss in a pressed mesocarp fibre and broken nuts in a pressed cake in a palm oil mill in real-time. The present invention is a predictive model or equation between a combination of a first input parameter, a second input parameter, a third input parameter, and a fourth input parameter, and it relates generally to a method of continuous, automated and integrated prediction of a first output and a second output in real-time. More specifically, the present invention pertains to a method for predicting the oil loss in a pressed mesocarp fibre and broken nuts found within a pressed cake in a palm oil mill. For instance, when the digester level is high (almost full), the digester drainage is continuously open, and the digester temperature is maintained between 85°C to 95°C, the prediction model using the machine learning algorithm is able to predict the OLWB to be in the range of 3.871% to 4.313% with cone press pressure at 80 bars. If the prediction method indicates that the OLWB is greater than 4.5%, the screw press speed can be adjusted, the digester level must be greater than 75% full, or the digester drainage must always be open, depending on the mill's process parameters. The OLWB% obtained will also enable us to ascertain the state of the fruitlets we received from the estates and to adjust the mill's processing guidelines to ensure OLWB less than 4.5% and broken nuts less than 20%. The current invention is a prediction model or equation technique using machine learning algorithm to ascertain the percentage of OLWB in the pressed mesocarp fibre and broken nuts (kernels) in the pressed cake of a pressed mesocarp fibre. The validation of the prediction method of the present invention against data obtained from laboratory analysis results in an acceptable error (measured in root mean squared error, RMSE) of 1%. The average value of OLWB falls within the range of 3.871% to 4.314% with a confidence interval of 95%. Analysis of variance results in a P-value of less than 0.05 when the screw press speed, digester levels, and cone press pressure are combined. This demonstrates that the oil loss in the press mesocarp fibre is greatly influenced by these input conditions. A machine learning algorithm's equation resulting from the interaction of the input parameters provides a low P-value. It is preferable for the digester's bottom drainage to always be open. The machine learning algorithm is retrained each time fresh oil palm fruitlets are brought to the oil mill for processing, allowing the customized equation to be developed to predict the OLWB% and broken nuts%. This confirms that these input parameters significantly contributes to the oil loss in the press mesocarp fibre. The interaction of the inputs parameters as provided by the equation generated by a machine learning algorithm which provides a low P-value. It is preferred for the bottom drainage of the digester to be continuously open. The customised equation generated allows for the prediction of the OLWB % and broken nuts % whereby the machine learning algorithm is retrained every time new oil palm fruitlets are brought to the oil palm mill for processing. A first object of the present invention is to provide a method for continuous, automated, and integrated prediction of a first output (OLWB %) and a second output (% of broken nuts) in real-time. The method is a predictive model or equation obtained from a machine learning algorithm between combinations and interactions of a first input parameter, a second input parameter, a third input parameter, and a fourth input parameter. A second objective of the present invention is to provide a continuous, automated, and integrated prediction method for oil loss in a pressed mesocarp fibre and broken nuts in a pressed cake in a palm oil mill in real-time. The prediction method uses an equation or predictive model between the combination and interaction of a first input parameter, a second input parameter, a third input parameter, and a fourth input parameter. A third object of the present invention is to provide a predictive model or equation between the input parameters of cone press pressure, screw press speed, digester level, and the opening and closing of a digester's bottom drainage. It is recommended for the digester's bottom drainage to remain continually open, and the temperature of the digester is fixed within an 85°C to 95°C range. A fourth object of the present invention is to provide a predictive model or equation for the OLWB in a pressed mesocarp fibre with a low P-value of less than 0.05 based on the interaction of the input parameters. A fifth object of the present invention is to provide a predictive model or equation for broken nuts in pressed cake with a low P-value of less than 0.05 based on the interaction of the input parameters. A sixth object of the present invention is to provide a root mean square error (RMSE) of the prediction model of less than 1% for the first output (OLWB %) in contrast to laboratory results. A seventh object of the present invention is to provide a root mean square error (RMSE) of the predictive model of less than 10% for the second output (broken nuts %) in comparison to laboratory results. An eighth object of the present invention is to provide an automated means in the palm oil milling process, in order to lessen reliance on people for monitoring tasks. The majority of mill procedures are carried out manually by mill operators at several stations, where factors like temperature, speed, pressure, and others are all manually changed depending on observations made by the mill operators. A ninth object of the present invention is to provide an automatic recording of all data and parameters in the multitude of storage of electronic devices, portable mobile devices, cloud computing networks, or any combination thereof in contrast to traditional methods, where data and parameters are manually recorded by the mill operators. While the present invention is described herein by way of example, it should be understood that the detailed description are not intended to limit the invention to the particular form disclosed but in contrary the invention is to cover all modifications, equivalents and alternatives falling within the scope of the present invention. It is important to note that the detailed explanation does not intend to restrict the invention solely to the depicted forms or drawings described. On the contrary, the intention is to encompass all possible modifications, equivalents, and alternatives that fall within the broad scope of the present invention. Terminology and phraseology used herein is solely used for descriptive purposes and is not intended as limiting in scope. The words such as “including”, “comprising”, “having”, “containing” or “involving” and other variations is intended to be broad and cover the subject matter as described including equivalents and additional subject matter not recited such as other components or steps. The present method provides a method for a continuous, automated and integrated prediction of a first output and a second output in real-time in a palm oil mill, whereby the method is a predictive model or equation between a first input parameter, a second input parameter, a third input parameter and a fourth input parameter. The probability value (P-value) of the predictive model is less than 0.05 for the first and second outputs. The confidence interval of the predictive model is at least 95% for the first and second outputs. The root mean square error (RMSE) of the predictive model for the first output is <1%. The root mean square error (RMSE) of the predictive model for the second output is less than 10%. The first output refers to oil loss on a wet basis in a pressed mesocarp fibre. The second output refers to broken nuts in a pressed cake of a pressed mesocarp fibre. A percentage of oil loss on a wet basis is calculated via a first predictive model or equation: (A1 – [A2*W] – [A3*X] – [A4*Y] – [A5*Z] + [A6*W*Y] – [A7*Y*Z]). Value of A1 is in a range of between 6.12846 to 6.50754, value of A2 is in a range of between 0.0078764 to 0.0083636, value of A3 is in a range of between 0.129301 to 0.137299, value of A4 is in a range of between 0.49761 to 0.52839, value of A5 is in a range of between 0.203215 to 0.215785, value of A6 is in a range of between 0.0068967 to 0.0073233 and value of A7 is in a range of between 0.085166 to 0.090434. The percentage of the broken nuts is calculated via a second predictive model or equation: (B1 + [B2*W] – [B3*X] – [B4*Y] + [B5*Y*X]). Value of B1 is in a range of between 9.118 to 9.682, value of B2 is in a range of between 0.216698 to 0.230102, value of B3 is in a range of between 0.123764 to 0.131419, value of B4 is in a range of between 20.3021 to 21.5579 and value of B5 is in a range of between 0.374386 to 0.397544. W is the first input parameter which refers to pressure of at least one cone press in a range of between 50 bars to 80 bars. X is the second input parameter which refers to speed of at least one cone press in a range of between 8 rpm to 10 rpm. Y is the third input parameter which refers to opening or closing of at least one bottom drainage of a digester. Z is the fourth input parameter which refers to level of at least one digester level. The percentage of the oil loss on a wet basis in a pressed mesocarp fibre is less than 4.50%. The root mean square error of the oil loss on a wet basis in a pressed mesocarp fibre is less than 1%. The percentage of broken nuts in the press cake of a mesocarp fibre is less than 20%. The root mean square error of the broken nuts in the press cake of a pressed mesocarp fibre is less than 10%. The invention is further described below: A method for continuous, automated and integrated prediction of a first output and a second output in real-time, whereby the method is a predictive model or equation generated by a machine learning algorithm which provides combinations and interactions between a first input parameter, a second input parameter, a third input parameter and a fourth input parameter. A first output refers to the percentage of oil loss in a pressed mesocarp fibre on a wet basis (OLWB). A second output refers to the percentage of broken nuts in a pressed cake of a pressed mesocarp fibre. The first, second, third and fourth input parameters refer to: - cone press pressure; - screw press speed; - opening or closing of bottom drainage; and - digester level. The goal is to use a prediction model or equation created by a machine learning algorithm to calculate the first and second outputs in real-time rather than manually or conventional methods. Every time fresh samples (oil palm FFB / fruitlets) are gathered from oil palm estates to be processed in the palm oil mill, the predictive model is retrained via the machine learning algorithm. This would enable effective processing of the oil palm fruitlets, enabling fast adjustment of the process conditions of the input parameters to guarantee OLWB is less than 4.5% and broken nuts is less than 0% at all times. Equation 1: A predictive method for continuous measurement of oil loss in a pressed mesocarp fibre, the method comprising calculating oil loss on a wet basis(OLWB) (%) with the following predictive model or equation generated by a machine learning algorithm while maintaining a digester temperature in a range of between 85°C to 95°C: OLWB (%) = (A1 – [A2*W] – [A3*X] – [A4*Y] – [A5*Z] + [A6*W*Y] – [A7*Y*Z]). OLWB (%) = (A1 – [A2*cone pressure] – [A3*screw press speed] – [A4*opening or closing of a bottom drainage] – [A5*digester level] + [A6*cone pressure*opening or closing of bottom drainage] – [A7*opening or closing of bottom drainage*digester level]) For the purposes of the present invention: - W refers to the cone press pressure; - X refers to the screw press speed; - Y refers to the opening or closing of bottom drainage; and - Z refers to digester level. The value of A1 is in a range of between 6.12846 to 6.50754. The value of A2 is in a range of between 0.0078764 to 0.0083636 The value of A3 is in a range of between 0.129301 to 0.137299. The value of A4 is in a range of between 0.49761 to 0.52839. The value of A5 is in a range of between 0.203215 to 0.215785. The value of A6 is in a range of between 0.0068967 to 0.0073233. The value of A7 is in a range of between 0.085166 to 0.090434. Equation 2: A prediction model or equation for continuous measurement of broken nuts in a pressed cake, the method comprising calculating broken nuts (%) with the following equation generated by a machine learning algorithm while maintaining a digester temperature in a range of between 85°C to 95°C: Broken nuts (%) = (B1 + [B2*W] – [B3*X] – [B4*Y] + [B5*Y*X]). Broken nuts (%) = (B1 + [B2*cone pressure] – [B3*screw press speed] – [B4*opening or closing of a bottom drainage] + [B5*opening or closing of a bottom drainage*screw press speed]) For the purposes of the present invention: - W refers to the cone press pressure; - X refers to the screw press speed; - Y refers to the opening or closing of bottom drainage; and - Z refers to digester level. The value of B1 is in a range of between 9.118 to 9.682. The value of B2 is in a range of between 0.216698 to 0.230102. The value of B3 is in a range of between 0.123764 to 0.131419. The value of B4 is in a range of between 20.3021 to 21.5579. The value of B5 is in a range of between 0.374386 to 0.397544. A method which is a predictive model or equation that uses combinations and interactions of a first input parameter, a second input parameter, a third input parameter, and a fourth parameter for a continuous, automated and integrated prediction of a first output and a second output in real-time. A method which is a predictive model or equation that uses combinations and interactions of a first input parameter, a second input parameter, a third input parameter, and a fourth parameter for a continuous, automated and integrated prediction of the oil loss in a pressed mesocarp fibre and broken nuts in a pressed cake in a palm oil mill in real-time. A method which is a predictive model or equation that uses combinations and interactions between the cone press pressure, screw press speed, opening or closing of bottom drainage and digester level for a continuous, automated and integrated prediction of the oil loss in a pressed mesocarp fibre and broken nuts in a pressed cake in a palm oil mill in real-time. When the null hypothesis of the statistical test is true, the P-value (probability) is a number used to describe or indicate how likely the data obtained would have occurred. This number will produce a statistical summary that is equal to or greater than the actual observed results, indicating how large or small the experimental error is. • The predictive model's probability value (P-value) for OLWB is less than 0.05. This probability value demonstrates that the input parameters - cone press pressure, screw press speed, opening or closing of bottom drainage, and digester level all contribute to oil loss in the pressed mesocarp fibre. Based on the predictive model, two combinations of the input parameters can be adjusted to achieve a P-value less than 0.05: a) the cone press pressure and the opening or closing of the bottom drainage, or b) the opening or closing of the bottom drainage and the digester level. • The predictive model's probability value (P-value) for broken nuts is less than 0.05. This probability value demonstrates that the input parameters - cone pressure, screw press speed and opening or closing of bottom drainage, and digester level all contribute to the broken nuts in the pressed cake of a pressed mesocarp fibre. Based on the predictive model, the screw press speed and digester level can be adjusted to ensure that the P-value for broken nuts is less than 0.05. The predictive model's confidence interval is at least 95% with OLWB less than 4.5%, ideally between 3.871 and 4.314%, for optimization with minimal OLWB in pressed mesocarp fibre. A confidence interval of at least 95% indicates that the predictive equation or model provides the needed accuracy or actual value. The predictive model's confidence interval is at least 95% with OLWB less than 4.5%, ideally between 3.871 and 4.314%, for optimization with minimal OLWB in pressed mesocarp fibre. A confidence interval of at least 95% indicates that the predictive equation or model provides the needed accuracy or actual value. The predictive model's confidence interval is at least 95% with broken nuts less than 20%, for optimization with minimal OLWB in pressed mesocarp fibre. A confidence interval of at least 95% indicates that the predictive equation or model provides the needed accuracy or actual value. EXAMPLES Example 1 Equation 1: The percentage of oil loss on a wet basis is calculated via a predictive equation: (A1 – [A2*W] – [A3*X] – [A4*Y] – [A5*Z] + [A6*W*Y] – [A7*Y*Z]). The validation of the prediction model or equation of the present invention against the laboratory analysis provides an acceptable error of <1% for OLWB The validation of the present invention's prediction model or equation against laboratory analysis yields an acceptable error of less than 1% for OLWB (i.e. the root mean square error (RMSE) of the predictive model is less than 1%). This demonstrates that the average difference between the anticipated and real values (wet lab analyses) is less than 1%, confirming the predictive model's or Equation 1's high accuracy. Equation 2: The percentage of the broken nuts is calculated via a predictive equation: (B1 + [B2*W] – [B3*X] – [B4*Y] + [B5*Y*X]). The validation of the prediction model or equation of the present invention against the laboratory analysis provides an acceptable error of <10% for broken nuts The validation of the present invention's prediction model or equation against laboratory analysis yields an acceptable error of less than 10% for broken nuts (i.e. the root mean square error (RMSE) of the predictive model is <10%). This demonstrates the prediction model's or Equation 2's excellent accuracy. The first input parameter, W means pressure of at least one cone press in a range of between 50 bars to 80 bars. The second input parameter, X means speed of at least one cone press in a range of between 8 rpm to 10 rpm. The third input parameter, Y denotes the opening or closing of at least one bottom drainage, with a value of 1 indicating ‘open’ and a value of 0 indicating ‘close’. It is preferable for the bottom drainage of the palm oil mill to be continuously open throughout the milling process. The fourth input parameter, Z means level of at least one digester which is preferably 75% full or most preferably >90% full. The temperature of at least one digester is in a range of between 85°C to 95°C. The present innovation allows you to change the input parameters to get the following results: • The percentage of oil loss on a wet basis less than 4.5%. • The percentage of broken nuts less than 20%. • RMSE of the predictive model is less than 1% for OLWB in comparison to the validation / analysis. • RMSE of the predictive model is less than 10% for broken nuts in comparison to the validation / analysis. Table 1: Validation between actual / lab analysis vs. the prediction model or Equation 1 for OLWB % Data A’ B’ A’ – B’ OLWB (%) OLWB (%) (%) (via the prediction (actual / lab analysis) equation) Data1 4.66 4.16 0.50 Data2 4.89 4.09 0.80 Data3 4.50 4.22 0.28 Data4 3.87 3.82 0.05 Data5 3.54 4.05 0.51 Data6 3.81 3.94 0.13 Data7 3.64 4.03 0.39 Data8 3.73 3.93 0.20 Data9 3.61 3.90 0.29 Data10 3.94 4.35 0.41 Data11 3.64 4.16 0.52 Data12 3.58 3.94 0.36 Data13 3.82 4.07 0.25 Data14 3.51 4.16 0.65 Data15 4.06 4.34 0.28 Data16 4.13 4.39 0.26 Data17 3.97 3.97 0.00 Data18 4.34 3.92 0.42 Data19 3.73 4.19 0.46 Data20 3.48 4.38 0.90 Data21 3.49 3.90 0.41 Data22 4.17 4.28 0.11 Data23 4.95 3.94 1.01 Data24 3.97 4.05 0.08 Data25 4.70 4.14 0.56 Data26 4.54 4.20 0.34 Data27 4.47 4.15 0.32 Data28 4.02 4.19 0.17 Data29 4.23 4.09 0.14 Data30 3.87 4.13 0.26 Data31 4.11 4.23 0.12 Data32 4.61 4.62 0.01 Data33 4.35 4.49 0.14 Data34 4.20 4.43 0.23 Data35 4.35 4.27 0.08 Data36 5.58 4.17 1.41 Data37 3.95 4.25 0.30 N 37 Mean 4.15 RMSE 0.46 Table 2: Validation between actual / lab analysis vs. the prediction model or Equation 2 for broken nuts % Data A B A – B Broken Nuts (%) Broken Nuts (%) (%) (via the prediction equation) (actual / lab analysis) Data1 4.35 25.23 20.88 Data2 3.97 28.24 24.27 Data3 3.32 29.34 26.02 Data4 10.22 13.44 3.22 Data5 11.19 8.89 2.30 Data6 11.31 8.23 3.08 Data7 5.83 9.47 3.64 Data8 2.59 7.71 5.12 Data9 13.29 21.79 8.50 Data10 10.65 13.78 3.13 Data11 8.29 11.44 3.15 Data12 13.10 33.65 20.55 Data13 8.67 4.81 3.86 Data14 4.00 4.55 0.55 Data15 2.35 0.00 2.35 Data16 0.35 0.00 0.35 Data17 -0.71 2.60 3.31 Data18 0.57 4.26 3.69 Data19 0.43 7.62 7.19 Data20 2.70 11.76 9.06 Data21 6.44 14.45 8.01 Data22 4.16 6.35 2.19 Data23 2.79 16.21 13.42 Data24 0.35 5.05 4.70 Data25 0.00 0.00 0.00 Data26 13.58 12.07 1.51 Data27 12.42 10.69 1.73 Data28 11.83 10.83 1.00 Data29 10.91 16.87 5.96 Data30 9.82 15.12 5.30 Data31 12.17 23.11 10.94 Data32 7.33 10.77 3.44 Data33 7.53 3.73 3.80 Data34 7.64 8.53 0.89 Data35 7.19 8.50 1.31 Data36 5.37 3.20 2.17 Data37 3.87 1.44 2.43 N 37 Mean 6.48 RMSE 8.94 The validation of the current invention's prediction method against data acquired from laboratory analysis yields an acceptable error of less than 1%, with the average value of OLWB lying between3.871% and 4.314% and a confidence interval of 95%. Using analysis of variance, a P-value of lessthan 0.05 is obtained for the cone press pressure, screw press speed, and digester levels. This demonstrates that these input parameters have a considerable impact on oil loss in the press mesocarp fibre. The interaction of input parameters as represented by an equation derived by a machine learning system provides a low P-value. The bottom drainage of the digester is preferred to be kept open at all times. The customized equation developed predicts the OLWB and broken nuts percentages, and the machine learning algorithm is retrained each time new oil palm fruitlets are supplied to the oil palm mill for processing. None of the previously mentioned prior arts specifically depict the method of the present invention. Furthermore, it is not readily apparent to experts in the relevant field how to arrive at the present invention simply by reading the prior art materials mentioned above. The inventors arrived at the method of the current invention through a series of extensive experimentation, testing, observations, and discussions utilizing their combined expertise and experience in this field. The specific parameters and / or combinations utilized in the method could not have been determined without significant effort, testing, analysis, or simply by reviewing prior art documents. As a result, there is a need in the art to provide a method per present invention. The inventors have derived the parameters for this invention through a series of meticulous trials, careful observations, and insightful discussions among experts in the field. Determining these parameters or their appropriate combination would have required significant effort, thorough testing, and detailed analysis, far beyond what could have been accomplished solely by reviewing existing documents in the field. Therefore, based on their knowledge, the inventors confidently claim that this invention is both novel and inventive.

Claims

CLAIMS 1. A method for a continuous, automated and integrated prediction of a first output and a second output in real-time in a palm oil mill, whereby the method is a predictive model or equation between a first input parameter, a second input parameter, a third input parameter and a fourth input parameter.

2. The method of Claim 1, wherein a probability value (P-value) of the predictive model is less than 0.05 for the first and second outputs.

3. The method of Claim 1, wherein a confidence interval of the predictive model is at least 95% for the first and second outputs.

4. The method of Claim 1, wherein a root mean square error (RMSE) of the predictive model for the first output is <1%. 5 The method of Claim 1, wherein a root mean square error (RMSE) of the predictive model for the second output is less than 10%.

6. The method of Claim 1, wherein the first output refers to oil loss on a wet basis in a pressed mesocarp fibre.

7. The method of Claim 1, wherein the second output refers to broken nuts in a pressed cake of a pressed mesocarp fibre.

8. The method of Claim 6, wherein a percentage of oil loss on a wet basis is calculated via a first predictive model or equation: (A1 – [A2*W] – [A3*X] – [A4*Y] – [A5*Z] + [A6*W*Y] – [A7*Y*Z]).

9. The method of Claim 8, wherein value of A1 is in a range of between 6.12846 to 6.50754, value of A2 is in a range of between 0.0078764 to 0.0083636, value of A3 is in a range of between 0.129301 to 0.137299, value of A4 is in a range of between 0.49761 to 0.52839, value of A5 is in a range of between 0.203215 to 0.215785, value of A6 is in a range of between 0.0068967 to 0.0073233 and value of A7 is in a range of between 0.085166 to 0.090434.

10. The method of Claim 7, wherein a percentage of the broken nuts is calculated via a second predictive model or equation: (B1 + [B2*W] – [B3*X] – [B4*Y] + [B5*Y*X]).

11. The method of Claim 10, wherein value of B1 is in a range of between 9.118 to 9.682, value of B2 is in a range of between 0.216698 to 0.230102, value of B3 is in a range of between 0.123764 to 0.131419, value of B4 is in a range of between 20.3021 to 21.5579 and value of B5 is in a range of between 0.374386 to 0.397544.

12. The method of Claims 8 and 10, wherein W is the first input parameter which refers to pressure of at least one cone press in a range of between 50 bars to 80 bars.

13. The method of Claims 8 and 10, wherein X is the second input parameter which refers to speed of at least one cone press in a range of between 8 rpm to 10 rpm.

14. The method of Claims 8 and 10, wherein Y is the third input parameter which refers to opening or closing of at least one bottom drainage of a digester. 15 The method of Claim 8, wherein Z is the fourth input parameter which refers to level of at least one digester level.

16. The method of Claim 8, wherein the percentage of the oil loss on a wet basis in a pressed mesocarp fibre is less than 4.50%.

17. The method of Claim 8, wherein root mean square error of the oil loss on a wet basis in a pressed mesocarp fibre is less than 1%.

18. The method of Claim 10, wherein the percentage of broken nuts in the press cake of a mesocarp fibre is less than 20%.

19. The method of Claim 10, wherein root mean square error of the broken nuts in the press cake of a pressed mesocarp fibre is less than 10%.