Average of L-2 norm of errors for 100 runs of each configuration \n
\r\n\t
",isbn:"978-1-83962-292-2",printIsbn:"978-1-83962-291-5",pdfIsbn:"978-1-83962-293-9",doi:null,price:0,priceEur:0,priceUsd:0,slug:null,numberOfPages:0,isOpenForSubmission:!1,hash:"5c4c0d41cf25d2e8fda944450ac46d95",bookSignature:"Prof. Constantin Volosencu",publishedDate:null,coverURL:"https://cdn.intechopen.com/books/images_new/9976.jpg",keywords:"Fuzzy Aggregation Operators, Fuzzy Set, Hesitant Fuzzy Set, Fuzzy Decision Making, Pythagorean Fuzzy Sets, Pythagorean Fuzzy Decision, Spherical Fuzzy Set, Linguistic Fuzzy Set, Fuzzy Control System, Fuzzy Expert Systems, Fuzzy Logic, Fuzzy Neural Network",numberOfDownloads:222,numberOfWosCitations:0,numberOfCrossrefCitations:0,numberOfDimensionsCitations:0,numberOfTotalCitations:0,isAvailableForWebshopOrdering:!0,dateEndFirstStepPublish:"June 4th 2020",dateEndSecondStepPublish:"September 10th 2020",dateEndThirdStepPublish:"November 9th 2020",dateEndFourthStepPublish:"January 28th 2021",dateEndFifthStepPublish:"March 29th 2021",remainingDaysToSecondStep:"6 months",secondStepPassed:!0,currentStepOfPublishingProcess:5,editedByType:null,kuFlag:!1,biosketch:"Dr. Volosencu is author of 10 books and has over 155 scientific papers published. He holds 27 patents and has developed electrical equipment for machine tools, spooling machines, high power ultrasound processes and other, with the homologation of 18 prototypes and 12 zero manufacturing series.",coeditorOneBiosketch:null,coeditorTwoBiosketch:null,coeditorThreeBiosketch:null,coeditorFourBiosketch:null,coeditorFiveBiosketch:null,editors:[{id:"1063",title:"Prof.",name:"Constantin",middleName:null,surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu",profilePictureURL:"https://mts.intechopen.com/storage/users/1063/images/system/1063.jpeg",biography:"Constantin Volosencu is a professor at the Polytechnic University of Timişoara, Department of Automation. He is the editor of 9 books, author of 10 books, 5 book chapters, and over 180 scientific papers published in journals and conference proceedings. He is also a holder of 27 patents, and a manager of research grants. He is a member of editorial boards of international journals, a former plenary speaker, a member of scientific committees, and chair at international conferences. His research is in the field of control systems, electrical drives, power ultrasounds, fuzzy logic, neural networks, fault detection and diagnosis, sensor networks, and distributed parameter systems. He has developed electrical equipment for machine tools, spooling machines, high-power ultrasound processes and others.",institutionString:"Polytechnic University of Timişoara",position:null,outsideEditionCount:0,totalCites:0,totalAuthoredChapters:"3",totalChapterViews:"0",totalEditedBooks:"8",institution:{name:"Polytechnic University of Timişoara",institutionURL:null,country:{name:"Romania"}}}],coeditorOne:null,coeditorTwo:null,coeditorThree:null,coeditorFour:null,coeditorFive:null,topics:[{id:"9",title:"Computer and Information Science",slug:"computer-and-information-science"}],chapters:[{id:"75188",title:"Location Selection for Smog Towers Using Zadeh’s Z-Numbers Integrated with WASPAS",slug:"location-selection-for-smog-towers-using-zadeh-s-z-numbers-integrated-with-waspas",totalDownloads:25,totalCrossrefCites:0,authors:[null]},{id:"75248",title:"Compensatory of Adaptive Neural Fuzzy Inference System",slug:"compensatory-of-adaptive-neural-fuzzy-inference-system",totalDownloads:16,totalCrossrefCites:0,authors:[null]},{id:"75189",title:"Fuzzy Modeling of Urban Water Supply Crisis",slug:"fuzzy-modeling-of-urban-water-supply-crisis",totalDownloads:17,totalCrossrefCites:0,authors:[null]},{id:"75035",title:"Towards a Fuzzy Context Logic",slug:"towards-a-fuzzy-context-logic",totalDownloads:24,totalCrossrefCites:0,authors:[null]},{id:"74830",title:"Fuzzy Multi-Attribute Decision Making (FMADM) Application on Decision Support Systems (SPK) to Diagnose a Type of Disease",slug:"fuzzy-multi-attribute-decision-making-fmadm-application-on-decision-support-systems-spk-to-diagnose-",totalDownloads:27,totalCrossrefCites:0,authors:[null]},{id:"74023",title:"Existence, Uniqueness and Approximate Solutions of Fuzzy Fractional Differential Equations",slug:"existence-uniqueness-and-approximate-solutions-of-fuzzy-fractional-differential-equations",totalDownloads:115,totalCrossrefCites:0,authors:[null]}],productType:{id:"1",title:"Edited Volume",chapterContentType:"chapter",authoredCaption:"Edited by"},personalPublishingAssistant:{id:"205697",firstName:"Kristina",lastName:"Kardum Cvitan",middleName:null,title:"Ms.",imageUrl:"https://mts.intechopen.com/storage/users/205697/images/5186_n.jpg",email:"kristina.k@intechopen.com",biography:"As an Author Service Manager my responsibilities include monitoring and facilitating all publishing activities for authors and editors. From chapter submission and review, to approval and revision, copyediting and design, until final publication, I work closely with authors and editors to ensure a simple and easy publishing process. I maintain constant and effective communication with authors, editors and reviewers, which allows for a level of personal support that enables contributors to fully commit and concentrate on the chapters they are writing, editing, or reviewing. I assist authors in the preparation of their full chapter submissions and track important deadlines and ensure they are met. I help to coordinate internal processes such as linguistic review, and monitor the technical aspects of the process. As an ASM I am also involved in the acquisition of editors. Whether that be identifying an exceptional author and proposing an editorship collaboration, or contacting researchers who would like the opportunity to work with IntechOpen, I establish and help manage author and editor acquisition and contact."}},relatedBooks:[{type:"book",id:"2020",title:"New Technologies",subtitle:"Trends, Innovations and Research",isOpenForSubmission:!1,hash:"170d84903f390df23023d0623d8577d3",slug:"new-technologies-trends-innovations-and-research",bookSignature:"Constantin Volosencu",coverURL:"https://cdn.intechopen.com/books/images_new/2020.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"2021",title:"Cutting Edge Research in New Technologies",subtitle:null,isOpenForSubmission:!1,hash:"5c14eed0ff55904ca388c886570c85fc",slug:"cutting-edge-research-in-new-technologies",bookSignature:"Constantin Volosencu",coverURL:"https://cdn.intechopen.com/books/images_new/2021.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"6024",title:"System Reliability",subtitle:null,isOpenForSubmission:!1,hash:"5cf0113f60979705f5b0b0ea0bac3028",slug:"system-reliability",bookSignature:"Constantin Volosencu",coverURL:"https://cdn.intechopen.com/books/images_new/6024.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"4646",title:"Cutting Edge Research in Technologies",subtitle:null,isOpenForSubmission:!1,hash:"a0c6cc73cb98936693e6e4845a19dfcf",slug:"cutting-edge-research-in-technologies",bookSignature:"Constantin Volosencu",coverURL:"https://cdn.intechopen.com/books/images_new/4646.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"7501",title:"Fault Detection and Diagnosis",subtitle:null,isOpenForSubmission:!1,hash:"5143fb77b96f488e4ec5dd6e7947904c",slug:"fault-detection-and-diagnosis",bookSignature:"Constantin Volosencu",coverURL:"https://cdn.intechopen.com/books/images_new/7501.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"9287",title:"Control Theory in Engineering",subtitle:null,isOpenForSubmission:!1,hash:"7c584de5f40193b636833aa812dab9d5",slug:"control-theory-in-engineering",bookSignature:"Constantin Volosencu, Ali Saghafinia, Xian Du and Sohom Chakrabarty",coverURL:"https://cdn.intechopen.com/books/images_new/9287.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"7656",title:"Fuzzy Logic",subtitle:null,isOpenForSubmission:!1,hash:"54f092d4ffe0abf5e4172a80025019bc",slug:"fuzzy-logic",bookSignature:"Constantin Volosencu",coverURL:"https://cdn.intechopen.com/books/images_new/7656.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"6603",title:"Actuators",subtitle:null,isOpenForSubmission:!1,hash:"33056f58590b5920dd938eff4810e8dc",slug:"actuators",bookSignature:"Constantin Volosencu",coverURL:"https://cdn.intechopen.com/books/images_new/6603.jpg",editedByType:"Edited by",editors:[{id:"1063",title:"Prof.",name:"Constantin",surname:"Volosencu",slug:"constantin-volosencu",fullName:"Constantin Volosencu"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"1591",title:"Infrared Spectroscopy",subtitle:"Materials Science, Engineering and Technology",isOpenForSubmission:!1,hash:"99b4b7b71a8caeb693ed762b40b017f4",slug:"infrared-spectroscopy-materials-science-engineering-and-technology",bookSignature:"Theophile Theophanides",coverURL:"https://cdn.intechopen.com/books/images_new/1591.jpg",editedByType:"Edited by",editors:[{id:"37194",title:"Dr.",name:"Theophanides",surname:"Theophile",slug:"theophanides-theophile",fullName:"Theophanides Theophile"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}},{type:"book",id:"3092",title:"Anopheles mosquitoes",subtitle:"New insights into malaria vectors",isOpenForSubmission:!1,hash:"c9e622485316d5e296288bf24d2b0d64",slug:"anopheles-mosquitoes-new-insights-into-malaria-vectors",bookSignature:"Sylvie Manguin",coverURL:"https://cdn.intechopen.com/books/images_new/3092.jpg",editedByType:"Edited by",editors:[{id:"50017",title:"Prof.",name:"Sylvie",surname:"Manguin",slug:"sylvie-manguin",fullName:"Sylvie Manguin"}],productType:{id:"1",chapterContentType:"chapter",authoredCaption:"Edited by"}}]},chapter:{item:{type:"chapter",id:"59091",title:"Efficient Matrix-Free Ensemble Kalman Filter Implementations: Accounting for Localization",doi:"10.5772/intechopen.72465",slug:"efficient-matrix-free-ensemble-kalman-filter-implementations-accounting-for-localization",body:'\nThe ensemble Kalman filter (EnKF) is a sequential Monte Carlo method for parameter and state estimation in highly nonlinear models. The popularity of the EnKF owes to its simple formulation and relatively easy implementation. In the EnKF, an ensemble of model realizations is utilized in order to, among other things, estimate the moments of the prior error distribution (forecast distribution). During this estimation and the assimilation of observations, many challenges are present in practice. Some of them are addressed in this chapter: the first one is related to the proper estimation of background error correlations, which has a high impact on the quality of the analysis corrections while the last one is related to the proper estimation of the posterior ensemble when the observation operator is nonlinear. In all cases, background error correlations are estimated based on the Bickel and Levina estimator [1].
\nWe want to estimate the state of a dynamical system \n
In the ensemble Kalman filter (EnKF) [2], an ensemble of model realizations
\nis utilized in order to estimate the moments of the background error distribution,
\nvia the empirical moments of the ensemble (1) and therefore,
\nand
\nwhere \n
A variety of EnKF implementations have been proposed in the current literature, most of them rely on basic assumptions over the moments of two probability distributions: the background (prior) and the analysis (posterior) distributions. In most of EnKF formulations, normal assumptions are done over both error distributions.
\nIn general, depending on how the assimilation process is performed, the EnKF implementation will fall in one of two categories: stochastic filter [3] or deterministic filter [4]. In the context of stochastic filters, for instance, the assimilation of observations can be performed by using any of the next formulations,
\nand
\nwhere the analysis covariance matrix \n
The forms of the Cholesky and the modified Cholesky decomposition were discussed by Golub and Van Loan in [5]. If \n
For a symmetric matrix \n
Recall \n
then, \n
In the ensemble Kalman filter based on a modified Cholesky decomposition (EnKF-MC) [6], the analysis ensemble is estimated by using the following equations,
\nand
\nIn this context, error correlations are estimated via a modified Cholesky decomposition. This provides an estimate of the inverse background error covariance matrix of the form,
\nor equivalently
\nwhere \n
Recall that the precision analysis covariance matrix reads,
\nmoreover, since \n
In the stochastic posterior ensemble Kalman filter (PEnKF-S) [7, 8], we want to estimate the moments of the analysis distribution,
\nbased on the background ensemble \n
where \n
where \n
where \n
We can use of Dolittle’s method in order to compute the factors \n
After some math simplifications, the next equations are obtained,
\nand
\nfor \n
Algorithm 1. Rank-one update for the factors \n
1. function UPD_CHOLESKY_FACTORS \n
2. Compute \n
3. for \n
4. Compute \n
5. Set \n
6. for \n
7. Compute \n
8. end for
\n9. end for
\n10. Set \n
11. return \n
12. end function
\nAlgorithm 1 can be used in order to update the factors of \n
Algorithm 2. Computing the factors \n
13. function COMPUTE_ANALYSIS_FACTOR \n
14. Set \n
15. for \n
16. Set \n
17. end for\n
\n18. return \n
19. end function\n
\nOnce the updating process has been performed, the resulting factors form an estimate of the inverse analysis covariance matrix,
\nFrom this covariance matrix, the posterior mode of the distribution can be approximated as follows:
\nwhere
\nwith \n
which can be utilized in order to build the analysis ensemble,
\nwhere \n
In addition, the columns of \n
The deterministic posterior ensemble Kalman filter (PEnKF-D) is a square root formulation of the PEnKF-S. The main difference between them lies on the use of synthetic data. In both methods, the matrices \n
In the PEnKF-D, the computation of \n
which are consistent with the dynamics of the numerical model, for instance, samples from (27) are driven by the physics and the dynamics of the numerical model. Finally, the analysis equation of the PEnKF-D
\nwhere
\nand
\nSince the moments of the posterior distribution are unchanged, we expect both methods PEnKF-S and PEnKF-D to perform equivalently in terms of errors during the estimation process.
\nDefinitely, the EnKF represents a breakthrough in the data assimilation context, perhaps its biggest appeal is that we can obtain a closed form expression for the analysis members. Nevertheless, as can be noted, during the derivation of the analysis equations, some constrains are imposed, for instance, the observation operator is assumed linear, and therefore, the likelihood \n
The prediction is obtained from the posterior distribution: \n
The information from numerical model is known like the prior of background: \n
The information from the observations is incorporated through the likelihood: \n
The posterior distribution is calculated: \n
If \n
A significant number of approaches have been proposed in the current literature in order to deal with these constraints; for instance, the particle filters (PFs) and the maximum likelihood ensemble filter (MLEF) are ensemble-based methods, which deal with nonlinear observation operators. Unfortunately, in the context of PF, its use is questionable under realistic weather forecast scenarios since the number of particles (ensemble members) increases exponentially regarding the number of components in the model state. Anyways, an extended analysis of these methods exceeds the scope of this document, but its exploration is highly recommended.
\nTaking samples directly from the posterior distribution is a strategy that can help to remove the bias induced by wrong assumptions on the posterior distribution (i.e., normal assumptions). We do not put the sights on finding the mode of the posterior distribution; instead of this, we want a set of state vectors that allow us to create a satisfactory representation of the posterior error distribution, then, based on these samples, it is possible to estimate moments of the posterior distribution from which the analysis ensemble can be obtained [9].
\nHereafter, we will construct a modification of the sequential scheme of data assimilation; first, we will describe how to compute the analysis members by using variations in Markov Chain Monte Carlo (MCMC)-based methods, and then, we will include the modified Cholesky decomposition in order to estimate a precision background covariance.
\nAn overview of the proposed method is as follows:
The forecast step is unchanged; the forecast ensemble is obtained by applying the model to the ensemble of states of the previous assimilation cycle.
The analysis step is modified, so that the analysis is not obtained anymore by, for instance (28), but we perform \nk\n iterations of an algorithm from the Markov Chain Monte Carlo (MCMC) family in order to obtain samples from the posterior distribution.
In order to be more specific in the explanation, let us define Metropolis-Hastings (MH) as the selected algorithm from the MCMC family. Now let us focus on MCMC in the most intuitive way possible. Let us define \n
The terms that involve \n
Initialize the Markov Chain, \n
Generate a candidate vector state \n
Obtain \n
If \n
Repeat Steps 2 through 4, for \n
Remove the first \n
The analysis is computed over the sample:
\nMCMC methods are straightforward to implement; when an enough number of iterations is performed, the behavior of the target function is captured on the chain [10]. This is true for even, complex density functions such as those with multiple modes. Briefly, let us focus on the fact that, generally, simulation-based methods such as MCMC explore a discretized grid, and as the mesh is refined, a huge number of iterations are needed before a high probability zone of the posterior distribution is reached [11]. Concretely, Hu et al. [12] proposed a family of modified MCMC dimension-independent algorithms under the name of preconditioned Crank-Nicolson (pCN) MCMC. These methods are robust regarding the curse of dimensionality in the statistical context. Initially, the Crank-Nicolson discretization is applied to a stochastic partial differential equation (SPDE) in order to obtain a new expression for the proposal distribution:
\nwhere \n
where \n
The procedure for the calculation of the analysis applying pCN Metropolis-Hastings is described as follows:
Initialize the Markov Chain, \n
Generate a candidate vector state using (3): \n
Obtain \n
If \n
Repeat Steps 2 through 4, for \n
Remove the first \n
The analysis is calculated over the sample: \n
Finally, in the data assimilation context, by using the modified Cholesky decomposition described in the earlier sections, the pCN-MH filter reads:
The forecast step remains unchanged; the forecast ensemble is obtained by applying the model to the ensemble of states of the previous iteration, but the estimation of background covariance matrix is calculated by modified Cholesky.
The update step is modified, so that the analysis is obtained by run k iterations of pCN-MH, to obtain a sample of the posterior error distribution.
Now we are ready to numerically test the methods discussed in this chapter.
\nWe assess the accuracy of the PEnKF-S and compare it against that of the LETFK implementation proposed by Hunt [13]. The numerical model is the Lorenz 96 model [11], which mimics the behavior of the atmosphere:
\nwhere \n
The experimental settings are described below:
An initial random solution \n
The reference solution is perturbed by using samples from a normal distribution with parameters \n
A similar procedure is performed in order to build a perturbed ensemble about \n
The assimilation windows consist of 15 equidistant observations. The frequency of observations is 0.5 time units, which represents 3.5 days in the atmosphere.
The dimension of the vector state is \n
The number of observed components is 50% of the dimension of the vector state.
Three ensemble sizes are tried during the experiments \n
As a measure of quality, the \n
A total of 100 runs are performed for each pair \n
The average of the error norms of each pair \n
\n\n | \n\nN\n | \nLETKF | \nPEnKF-S | \n
---|---|---|---|
0.05 | \n20 | \n22,6166 | \n21,2591 | \n
40 | \n20,5671 | \n18,2548 | \n|
60 | \n20,0567 | \n17,8824 | \n|
0.10 | \n20 | \n23,1742 | \n21,0725 | \n
40 | \n20,9513 | \n18,3542 | \n|
60 | \n18,5048 | \n17,8240 | \n|
0.15 | \n20 | \n24,8201 | \n20,9059 | \n
40 | \n21,1314 | \n18,1731 | \n|
60 | \n20,8487 | \n17,7590 | \n
Average of L-2 norm of errors for 100 runs of each configuration \n
Some plots of the \n
Local ensemble transform Kalmar filter (LETKF).
Holding the settings from the previous section, experiments are performed by using the PEnKF-D. The results have similar behavior to that of the LETFK and the PEnKF-S implementation proposed by Niño et al. in [8]. This can be appreciated in \nFigures 1\n–\n3\n. As can be seen, the error distributions reveal similar behavior across all compared filters.
\nPosterior ensemble Kalmar filter deterministic (PEnKF-D).
Posterior ensemble Kalmar filter stochastic (PEnKF-S).
We will describe experiments of our proposed filter based on a modified Cholesky decomposition and the preconditioned Crank-Nicolson Metropolis-Hastings. Again, the numerical model is the Lorenz 96 model. In this section, we are mainly interested on describing the behavior of the method, especially when the observational operator is nonlinear.
\nThe experimental settings are as follows:
The observational operator is quadratic: \n
The vector of states has \n
The ensemble size is \n
The length of the Markov Chain is 1000.
The proposal is pCN with \n
\n\nFigure 4\n describes the distribution of six of the 40 components of the state vector after 1000 iterations of the algorithm, attempting to visualize if the actual value is within or near the region of the highest probability density described by the sample contained in the Markov Chain.
\nHistogram of six components of the state vector after 1000 iterations of the algorithm, the dashed lines indicate the position of the true value for the respective component.
Note that, not in all cases, the actual values of the model components were located at the peaks of the histogram, but most of them were within or near to zones of high probability described by the sample. This result is important if we take into account that probably the posterior distribution is not normal as is the case of quadratic observation operators.
\nIn this chapter, efficient EnKF implementations were discussed. All of them are based on a modified Cholesky decomposition wherein a precision background covariance is obtained in terms of Cholesky factors. In the first filter, the PEnKF-S, synthetic data are utilized in order to compute the posterior members, as done in stochastic formulations of the filter. Even more, a sequence of rank-one updates can be applied over the factors of the prior precision matrix in order to estimate those of the posterior precision. In the second filter, the PEnKF-D, synthetic data are avoided by using perturbations obtained from the physics and the dynamics of the numerical model. Finally, a MCMC-based filter is obtained in order to reduce the impact of bias when Gaussian assumptions are broken during the assimilation of observations, for instance, when the observation operator is nonlinear. Numerical experiments with the Lorenz 96 model reveal that the proposed filters are comparable to filters from the specialized literature.
\nThis work was supported by the Applied Math and Computational Science Lab (AML-CS) at Universidad del Norte in Barranquilla, Colombia.
\nDue to their high nutrients contents, soybean meal and yellow corn are conventional feedstuffs in poultry feeds. Moreover, these two feed ingredients are also high in demand by other animals (soybean meal) and humans (yellow corn). The global consumption of poultry products, such as meat or eggs, appears to be increasing in the developing countries. Therefore, the global demands of the main poultry feedstuffs would increase leading to higher cost of poultry production. Studies have shown that the world’s population is expected to increase to 9.1 billion by the year 2050, [1]. This tremendous increase in population would produce competition in the available poultry feed ingredients for poultry nutrition. Furthermore, this increase in population will increase demand for poultry products. As a result, the availability of feed ingredients for poultry nutrition would become more competitive. In addition, there is an increasing trend to produce biofuel from feedstuffs, especially corn, to meet the demand all over the world. This further poses a serious food security risk, especially in the developing countries.
Currently, efforts are being made worldwide to use alternative sources of protein and energy to be substituted for soybean meal and yellow corn in monogastric animals such as poultry and swine. It is known that some developing countries produce a huge amount of alternative feedstuffs, considered as agro waste by–products such as wheat bran, rice bran, cotton seed meal, copra meal and palm kernel cake. However, many of these agro waste by–products are featuring on presence of non-starch polysaccharides (NSPs) such as xylan and mannan as well as anti-nutritional factors [2].
The NSPs are found to be the main reason for increasing the viscosity in the small intestine of the birds, and hence lead to increased moisture content of the excreta. Hence, the productivity and health status of the chickens could be affected [2]. Therefore, the inclusion of these agro waste by–products in poultry feed are limited. The nontraditional feedstuffs can be defined as those feed ingredients that have not been conventionally or commercially used in poultry rations. This chapter discusses the nontraditional feedstuffs with potential to be replaced partially or totally with soybean meal and yellow corn in poultry feeds.
It is well known that yellow corn is used as a main source of energy ingredient in poultry diets [2]. There are some nonconventional feed ingredients that can substitute certain amount of yellow corn in poultry rations. However, there are some limitations such as presence of anti-nutritional factors that lead to decrease feed intake and growth performance (Table 1). The other important point to consider is that the lack of knowledge about the composition of nutrients and their availability, due to the lack of research centers in the developing countries limit use of these feed ingredients.
Sorghum is the main food grain in Africa and parts of India and China [3]. The nutritive value of sorghum is almost 90–95% similar to that of yellow corn. Moreover, its global price is less than yellow corn [4]. The problem of sorghum is the high tannins content, which is water soluble polyphenolic metabolites and leads to reduce growth performance of poultry. Tannins in higher concentration are anti-nutritional because made chelates and reduce protein digestibility [5]. Sorghum is usually classified as bird resistant (less than 0.5% tannin) or non-bird resistant (1.5% tannin or higher) varieties. The negative effects of tannins are decreasing growth, feed intake, protein digestibility, egg production and leg abnormalities with broilers [4]. There are some procedures that can be applied to the sorghum to minimize tannins and improve the nutritive value of such feed ingredients. These methods include soaking in alkali solution and water. It is reported that tannic acid can be hydrolyzed in the chicks to gallic acid which excreted in urine as 4 – O – methyl gallate [4]. Therefore, the action of methyl donors such as calcium hydroxide or slurry of sodium carbonate could be included in poultry rations to improve the feed intake of high tannin sorghum. As a result, low tannin sorghum can completely replace yellow corn in poultry diets.
Wheat bran is the outer seed coat from flour mills. High in fiber, low in metabolizable energy (ME) and its usage in poultry nutrition is limited [4]. The ME can be increased up to 10% by simple steam pelleting, and the availability of phosphorus up to 20% under the same condition [6]. This by product could be beneficial for gut health which is reported to modify the gut microflora [4]. It is reported that wheat bran can be added in poultry diets up to 5–8% without negative effect [4]. Wheat bran contains xylan which may lead to increase viscosity in the small intestines. Therefore, xylanase supplementation is recommended for broilers fed more than 15% wheat bran in their diets [4].
Alternative and clean sources of energy are more attractive nowadays against fossil energy. The production of biofuel has globally increasing [7]. Therefore, the by-product obtained from this process is known as distillers dried grain with solubles (DDGS). It can be defined as a product obtained after ethanol extraction by distillation from the yeast fermentation, and drying at 75% of the resultant [8]. Including DDGS in poultry diets to replace part of yellow corn and soybean meal have shown positive results in terms of growth performance [9]. The main limitation of using DDGS in monogastrics is the variability of its nutrients content and availability [9]. This is due to the variation of growing conditions, ethanol production method and oil extraction. Therefore, it was reported that there are two types of DDGS; high protein and conventional DDGS (Table 2).
Ingredient | Limitation |
---|---|
Sorghum | High tannins content. |
Wheat bran | High fiber content, low metabolizable energy. |
Distillers dried grains with solubles (DDGS) | Variability and availability of nutrients. |
Date wastes | High fiber content in the date pits, low lysine, methionine, leucine and isoleucine. |
Millets | High fiber and tannins. |
Alternative energy sources that can replace yellow corn in poultry diets.
Nutrient | High protein DDGS | Conventional DDGS |
---|---|---|
Dry matter | 83.10 | 89.80 |
Crude protein | 34.10 | 27.10 |
Crude fiber | 8.35 | 7.85 |
Ether extract | 7.91 | 9.63 |
Arginine | 1.49 | 1.10 |
Cystine | 0.58 | 0.45 |
Glycine | 1.25 | 0.60 |
Histidine | 0.88 | 0.62 |
Isoleucine | 1.26 | 1.15 |
Leucine | 4.32 | 2.40 |
Lysine | 1.16 | 0.70 |
Methionine | 0.74 | 0.50 |
Phenyl alanine | 1.57 | 1.35 |
Serine | 1.60 | 1.30 |
Threonine | 1.31 | 0.93 |
Tryptophan | 0.30 | 0.20 |
Tyrosine | 1.34 | 0.80 |
Valine | 1.60 | 1.40 |
Metabolizable energy (Kcal/Kg) | 2628 | 2628 |
Nutrient composition of DDGS (% as –fed basis) [9].
Not only can DDGS provide energy in poultry diets, but also can provide protein and available phosphorus. It was shown that DDGS can be included in broiler diets at 8% or 15% in starter and grower phase, respectively without negative effects in their performance [8]. The supplementation of fiber-degrading enzyme could be an efficient way to enable the use of increased levels of DDGS in poultry and pig diets [10].
Dates are rich in vitamins and minerals. Usually, dates wastes consisting on the pulp and pits (stones). Date wastes are high in fiber, low in lysine, methionine, leucine and isoleusine [11]. The limitation of using date wastes is the high crude fiber in the date pits. Date wastes can be included in poultry diets up to 30% without negative effects on their performance [12]. In addition, the use of 30% of date pits (stones) with a supplementation of multi enzymes in broiler diets had no adverse effects on the final body weight [13]. Regarding date pits meal, it could be fed to laying hens up to 5% without adverse effects on their performance and egg quality. In addition, broilers fed diet incorporated with 4% date pits meal showed an ability to resist the deleterious effects of aflatoxine B1 [14].
Millets is adrought-resistant plant that produces a nutritious grain. It can be grown successfully under environmental conditions where corn and wheat fail to survive [15]. The nutrient content is variable, so that it contains 8–10% CP, 3395–3738 kcal/kg metabolizable energy, 3.60–5.27% fat and 1.59–2.36% fiber [15]. The limitation of using high levels of millets in poultry diets is the tannin content and fiber [16].
Routinely, soybean meal is used as a main source of protein ingredient in poultry diets [4]. There are some nontraditional feed ingredients that can replace certain amount of soybean meal in poultry diets. Nevertheless, there are some limitations such as presence of anti-nutritional factors that lead to reduce feed intake and growth performance (Table 3).
Canola crop is growing widely in the west of Canada as well as in other parts of the world [4]. The production of canola was influenced by the increasing demand for canola oil. Canola meal is the by-product of oil extraction, and lysine content is less than that of soybean meal. However, sulfur-containing amino acids are higher than that of soybean meal.
The problem of using canola meal in poultry feeds is the presence of glucosinolates, senapine, phytate, fibers, tannins as well as it has low metabolizable energy [17]. It was found that feeding canola meal to layers led to the occurring of fishy taint in egg and the reduction egg size [4].
There are attempts to improve the nutritional quality of canola meal by extrusion or solid-state fermentation using lactic acid bacteria [6, 18]. Therefore, it was reported that canola meal can be incorporated in poultry diets up to 5–8% [4], or up to 10% in broilers fed fermented canola meal based diet [17].
Peanut meal is a by-product from oil extraction. It contains 0.5–1% oil and 47% CP. The problem of using peanut meal in poultry diets is the trypsin inhibitors. Fortunately, it can be detoxifying by heat treatment during oil extraction. The issue to consider is that its potential aflatoxin contamination. To overcome this problem, the feedstuff could be supplemented with sodium-calcium aluminosilicates because these minerals bind with aflatoxin preventing its absorption [4].
Peas can be used in poultry diets depending on local economic conditions. It contains moderate amount of energy and protein. The limitation to use peas in poultry rations is the lack of sulfur containing amino acids, and moderate energy levels [4].
The use of low alkaloid lupins in poultry diets is going to be increased in certain regions of the world [4]. The high level of fiber in the seed leads in low metabolizable energy compared to soybean meal. Although lupins are much lower in methionine and lysine, many reports suggested that sweet lupins are comparable to soybean meal in terms of protein quality [4].
Sesame meal is very deficient in available lysine. It contains high level of phytate which may cause problems with calcium absorption. Therefore, skeletal disorders or poor egg shell quality in laying hens may be occurred. It contains 35.1–47% CP [16]. It is recommended that diet incorporated with more than 10% sesame meal should be increased by 0.2% extra calcium [4].
Blood meal is high in protein (65–85%), rich in lysine, arginine, methionine, cysteine and leucine. However, it is very poor in isoleucine [19]. The use of blood meal is very limited in poultry diets because of its palatability and poor growth rate [4]. It was reported that blood meal can be incorporated up to 3% in broiler diets without negative effects in their performance [19].
Tropical regions have an abundant amount of palm kernel cake (PKC), which is considered an agro-industrial waste derived from the extraction process of oil from palm fruits. It has been used in poultry diets as an alternative to soybean meal. Nevertheless, the use of PKC is limited in monogastrics because of its high content of fibers, coarse texture, and non-starch polysaccharides (NSPs) [2, 20, 21, 22, 23, 24]. The main NSPs in the PKC are mannan, xylan, arabinoxylan, and glucoronoxylan [20]. This is considered a significant issue faced by nutritionists, and it has limited the use of PKC for manipulation of feed formulation. It has been reported that 10% is the maximum level of PKC that can be given to broiler chickens. However, solid-state fermentation by cellulolytic bacteria may improve the nutritive value of PKC to be incorporated up to 15% in the diet [2, 24].
The treated PKC by enzyme [25], cellulolytic bacteria via solid state fermentation [2, 23, 24] or extrusion [26] may contribute to improve the nutritive value and poultry performance (Table 4). It was reported that extrusion led to 6% increase in apparent metabolizable energy and 32% in crude protein digestibility in broiler chickens [27].
Ingredient | Limitation |
---|---|
Canola meal | Presence of glucosinolates, senapine, phytate, fibers, tannins, and low metabolizable energy. |
Peanut (groundnut) meal | Trypsin inhibitors, potential aflatoxin contamination. |
Peas | Lack of sulfur containing amino acids, and moderate energy levels |
Lupins | High fiber, low metabolizable energy. |
Sesame meal | High levels of phytate. |
Blood meal | Palatability and low growth rate. |
Palm kernel meal | High fiber, coarse texture and high NSPs. |
Cottonseed meal | High fiber, gossypol, dry and dusty nature, phytate, sterculic acid. |
Feather meal | Low in amino acids availability. |
Insects and worms | Microbial deterioration and lipid oxidation during storage. |
Earthworms | High fat (PUFA), and lipid oxidation during storage. |
Algae | High fat (PUFA), and lipid oxidation during storage. |
Azolla | High fiber content. |
Single – cell protein | High fat (PUFA), and lipid oxidation during storage. |
Alternative protein sources that can replace soybean meal in poultry diets.
Nutrient (%) | PKC [23] | FPKCa1 [23] | FPKCb2 [23] | PKC [26] | Extruded PKC [26] |
---|---|---|---|---|---|
Crude protein | 16.43 | 16.80 | 16.68 | 16.90 | 16.90 |
Dry matter | 91.42 | 92.62 | 92.44 | 89.81 | 91.79 |
Ash | 474 | 4.67 | 4.80 | 4.50 | 5.70 |
Crude fiber | 16.96 | 14.09 | 14.29 | 17.30 | 14.60 |
NDF | 82.29 | 71.70 | 73.54 | 75.00 | 75.40 |
ADF | 51.48 | 47.27 | 47.45 | 37.30 | 39.30 |
Indispensable amino acids | |||||
Lysine | 0.37 | 0.41 | 0.38 | 0.5 | 0.46 |
Leucine | 0.89 | 0.94 | 0.95 | 1.08 | 1.05 |
Isoleucine | 0.50 | 0.59 | 0.53 | 0.60 | 0.55 |
Valine | 0.69 | 0.78 | 0.72 | 0.90 | 0.87 |
Phenyl alanine | 0.57 | 0.66 | 0.63 | 0.66 | 0.57 |
Threonine | 0.41 | 0.51 | 0.46 | 0.54 | 0.50 |
Histidine | 0.23 | 0.29 | 0.24 | 0.31 | 0.31 |
Methionine | 0.22 | 0.27 | 0.26 | 0.30 | 0.28 |
Arginine | 1.60 | 1.76 | 1.69 | 1.94 | 1.95 |
Glycine | 0.60 | 0.78 | 0.71 | 0.80 | 0.81 |
Dispensable amino acids | |||||
Aspartic acid | 1.12 | 1.27 | 1.23 | 1.14 | 1.15 |
Glutamic acid | 2.48 | 2.80 | 2.76 | 3.06 | 3.17 |
Proline | 0.44 | 0.59 | 0.52 | 0.57 | 0.53 |
Serine | 0.56 | 0.69 | 0.66 | 0.75 | 0.74 |
Tyrosine | 0.25 | 0.24 | 0.24 | 0.30 | 0.31 |
Cysteine | 0.20 | 0.22 | 0.21 | 0.36 | 0.17 |
Alanine | 0.62 | 0.70 | 0.71 | 0.87 | 1.10 |
Nutrient content of palm kernel cake and treated palm kernel cake (dry matter basis).
FPKCa; fermented palm kernel cake by P. polymyxa ATCC 842.
FPKCb; fermented palm kernel cake by P. curdlanolyticus DSMZ 10248.
Cottonseed meal is a byproduct after oil extraction. Usually, this byproduct used for poultry in cottonseed producing regions [4]. It is high in crude protein (41%). However, the big problem for using cottonseed meal in poultry rations are the high fiber levels (14.5%) and gossypol [4]. Gossypol is a yellow polyphenolic pigment, and usually found at 0.1% free gossypol. The big issue with gossypol is binding with lysine during processing, and then the lysine will be unavailable to the chickens. The byproduct is not acceptable by poultry because of its dry and dusty nature [3]. Gossypol may lead to decrease feed intake and growth rate in broiler chickens [3]. The byproduct is low in calcium, and the phosphorus is chelated with phytate. Therefore, phytase supplementation could be beneficial to release unavailable phosphorus. In case cottonseed meal is used for poultry, it is recommended to supply fish meal to balance the essential amino acids and calcium [3].
The other important point to consider with gossypol is that it leads to discoloration of the yolk in laying hens. It causes a olive-green color in the yolk, especially during egg storage at low temperature [3, 4]. The other problem with cottonseed meal is the presence of sterculic acid witch found to cause a pink color in the albumen. However, this can be avoided by using a byproduct with less residual oil because of the content of cyclopropenoid fatty acids [5].
It has been found that iron can bind with gossypol by 1:1 ratio, and may detoxify the gossypol. Therefore, the addition of 0.5 kg ferrous sulfate/tonne allowed the broilers and layers to tolerate up to 200 ppm and 30 ppm free gossypol, respectively without any negative effect in their performance [4].
In case iron was supplemented to cottonseed meal based diet, the balance between iron and copper should be considered to be 10: 1 iron to copper, respectively.
Studies have also shown that enzyme supplementation (β-glucanase and xylanase) may lead to increase the metabolizable energy and protein utilization in broiler chickens [28].
Feathers are considering as an industrial waste resulted during birds processing in slaughter houses. Several million tons of feathers are generated from the poultry processing industry are disposed as a waste [29, 30]. Feather meal contains about 85% crude protein, 5% cysteine and 3000 kcal/kg metabolizable energy. The cysteine availability is about 60% depending on the processing conditions [4].
Usually, feathers are partially dried, and hence steam-treated to introduce hydrolysis. However, the extreme temperature will lead to destruct the amino acids, especially lysine. Therefore, leads to reduce the amino acids digestibility. To overcome this problem, the use of keratinase enzyme may play an important role in improving the protein digestibility [29] and poultry performance [4]. In addition, fermentation with bacteria-degrading keratin such as Bacillus licheniformis for five days at 50°C can produces a fermented product comparable in nutritional value to soybean meal [4].
Some reports mentioned that B. subtilis and Aspergillus fumigatus had an ability to degrade keratin in feathers [30]. Feather meal can be included in poultry diets at 2–3%. Nevertheless, the fermented feather meal may give promised results in poultry nutrition, and therefore it would be an additional commercial benefit for the poultry industry by replacing part of soybean meal in poultry feeds.
Insects can be used to produce cheap source of protein. It is known that insects are considered as a natural food for birds. Insects are rich in protein (40–76%) and essential amino acids [31], particularly sulfur containing amino acids [32]. Insects meal are usually featuring on high fat content [31]. Therefore, microbial deterioration and lipid oxidation should be considered during storage [33]. Ssepuuya et al. [34] indicated that insects meal may replace the conventional protein sources by 10–100% without any negative growth performance whether in fish or poultry. It was also mentioned by Kareem et al. [31] that the incorporation of black soldier fly larvae to broiler diets up to 10% had no negative effect in their growth performance under humid tropical environment. In addition, no adverse effects on growth performance, carcass characteristics, hematological and serum biochemical indices in growing Japanese quail when meat and bone meal replaced with Spodoptera littorails in their diets [35]. It was claimed by Neumann et al. [36] that partly adding defatted insects meal of Hermetia illucens larvae in broiler diets – 26% and 22% in starter and grower phase, respectively – were acceptable. In terms of meat quality, it was reported that complete substitution of soybean meal by Hermetia illucens led to inducing lipid oxidation in broiler meat [37]. This was attributed to the high content of poly unsaturated fatty acids (PUFA) in Hermetia illucens.
Earthworms are a natural source of protein for poultry raised in free-range system. Earthworm can produced even in small-scale system. Earthworms species require a temperature ranging from 15 to 25°C, and 60–85% soil moisture content [38]. It can be considered as an alternative source of protein (64–76%) [39]. At the same time, it can be degrade animal manure to clean the environment. It was reported that the total essential amino acids in earthworms are comparable with egg protein. Moreover, the omega – 3 PUFA are quite high and similar to that of some fish oil [40]. It was mentioned by Parolini et al. [38] that earthworms contain 6–11% fat, 5–21% carbohydrate, 2–3% minerals and range of vitamins, especially niacin and cyanocobalamin. In comparison with insects meal, it has been found that earthworm meal has no deficiencies in the essential amino acids and better fatty acids profile with no chitin content, so that it was more acceptable and palatable for chickens [38]. Earthworm meal could be integrated in broiler diets up to 10% without negative effects in growth performance and meat quality [38].
Algae represent an important source of unconventional protein (50–60%), oils, vitamins, minerals, antioxidant and colorants [41], carotenoids, omega-3 and omega-6 PUFA [42, 43]. Some types of algae contain up to 76% crude protein [44]. In terms of nutrition, algae were used in broiler diets up to 16% without adverse effects. On the other hand, it was a replacement for approximately 60% of soybean meal and 40% of animal vegetable blended fat into practical broiler diets [44].
The most common species of algae used in poultry nutrition are chlorella and Spirulina. It was reported by Moury et al. [45] that supplementation of Spirulina platensis in broiler diets may completely replace the incorporation of vitamin-mineral premix. Moreover, it can be substitute the antibiotic usage in animals [46].
It is reported that algae can be a good option for 100% organic poultry feed [47]. Neumann et al. [36] reported that incorporation of Spirulina platensis at 21% and 17% in starter and grower phase, respectively was acceptable. However, nutritionists have to pay attention to the presence of PUFA in algae which may affect the meat quality of broilers and lead to lipid oxidation. Gkarane et al. [37] mentioned that complete substitution of soybean meal in broiler diets by Arthrospira platensis influenced the meat quality and led to lipid oxidation.
Azolla is an aquatic and floating fern of the family Azollaceae. It contains 25–35% crude protein, 10–15% minerals and 7–10% amino acids, especially lysine [48]. Azolla forms a symbiotic with blue green algae which lives within its leaves. This relationship makes azolla as a beneficial source of protein, and can be fed safely to the farm animals [49]. It is recommended that azolla (Azolla pinnata) can be incorporated in poultry diet up to 5% with positive effect on their growth performance [49]. The limitation of using high levels of Azolla is its high level content [48].
The production of single-cell protein (SCP) can be done by microbial fermentation with selected strains of microorganisms. SCP also known as microbial protein or bio-protein [50]. Bacteria such as Pseudomonas spp. can be grown in methanol, ethanol and organic acids [3]. The protein and sulfur containing amino acids in bacteria are higher than that of yeast. The oil content in bacteria and yeast is high and rich in unsaturated fatty acids. Chen et al. [51] concluded that SCP produced by Chlostridium autoethanogenum had 88.93% crude protein and most of essential amino acids were higher than that of fish meal.
The incorporation of 15% of SCP in pigs diet exhibited a comparable results with those group of pigs fed diet containing soybean meal [3]. It is recommended that SCP can be included in 2–5% in broiler diets, and up to 10% in laying hens [3].
It is known that – ingredients mentioned above – insects, worms, earthworms, algae, azolla and SCP contain significant amount of oil. In addition, these ingredients can provide omega-3 and omega-6 PUFA to the poultry [42, 43]. Interestingly, these ingredients are rich in vitamins and minerals as mentioned above [46].
In conclusion, the use of alternative feedstuffs nowadays in poultry sector is going to be increased because of their nutritive quality and as a cheap source of protein and energy. In addition, these nontraditional feedstuffs are not competitive with humans. At the same time, their inclusion to poultry diets can replace portions of soybean meal and yellow corn. Therefore, reduce the cost of production.
Open Access publishing helps remove barriers and allows everyone to access valuable information, but article and book processing charges also exclude talented authors and editors who can’t afford to pay. The goal of our Women in Science program is to charge zero APCs, so none of our authors or editors have to pay for publication.
",metaTitle:"What Does It Cost?",metaDescription:"Open Access publishing helps remove barriers and allows everyone to access valuable information, but article and book processing charges also exclude talented authors and editors who can’t afford to pay. The goal of our Women in Science program is to charge zero APCs, so none of our authors or editors have to pay for publication.",metaKeywords:null,canonicalURL:null,contentRaw:'[{"type":"htmlEditorComponent","content":"We are currently in the process of collecting sponsorship. If you have any ideas or would like to help sponsor this ambitious program, we’d love to hear from you. Contact us at info@intechopen.com.
\\n\\nAll of our IntechOpen sponsors are in good company! The research in past IntechOpen books and chapters have been funded by:
\\n\\nWe are currently in the process of collecting sponsorship. If you have any ideas or would like to help sponsor this ambitious program, we’d love to hear from you. Contact us at info@intechopen.com.
\n\nAll of our IntechOpen sponsors are in good company! The research in past IntechOpen books and chapters have been funded by:
\n\n