CN103218748A - Diagnostic method of vegetable diseases and portable system - Google Patents
Diagnostic method of vegetable diseases and portable system Download PDFInfo
- Publication number
- CN103218748A CN103218748A CN2013100844710A CN201310084471A CN103218748A CN 103218748 A CN103218748 A CN 103218748A CN 2013100844710 A CN2013100844710 A CN 2013100844710A CN 201310084471 A CN201310084471 A CN 201310084471A CN 103218748 A CN103218748 A CN 103218748A
- Authority
- CN
- China
- Prior art keywords
- symptom
- sample
- vector
- disease
- input
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses a diagnostic method of vegetable diseases and a portable system. The diagnostic method comprises the steps of establishing a vegetable disease diagnostic knowledge fuzzy set and a symptom importance membership degree to vegetable plants which have pathological manifestations in appearance; establishing an input vector based on a common semantic space according to the symptom importance membership degree and symptom characteristics; based on the established input vector, establishing a fuzzy neural network based on an improved genetic algorithm, and conducting training to obtain a mature fuzzy neural network; and inputting vegetable disease symptoms into the mature fuzzy neural network to obtain an output vector, and determining a diagnosis result. According to the diagnostic method of the vegetable diseases and the portable system, the input vector establishing method based on the common semantic space is used for establishing, bearing and describing the input vector of the symptom characteristics and symptom importance, the problems of being poor in representation of a sample set, multiple in contradiction samples, not obvious in embodying of diagnostic rules and the like are solved, an improved fuzzy neural network model of the genetic algorithm is used in a combined mode, so that the accuracy of the vegetable disease diagnosis result is high, and fault-tolerant inferential capacity is strong.
Description
Technical field
The present invention relates to the Agricultural Information technical field, relate in particular to a kind of vegetable disease diagnostic method and portable system.
Background technology
In the agricultural research field, realize the vegetable disease diagnosis at present, for example the use of image processing techniques, expert system analysis and fuzzy neural network etc. by multiple technologies and method.
Plant disease diagnostic method based on graphical analysis, it develops crops disease diagnosing system based on Computer Image Processing, this system utilizes the corps diseases information database under the support of video camera, image pick-up card and light-source box, to the disease image handle, identification, expert diagnosis.But the subject matter that the plant disease diagnostic method that is based on graphical analysis exists is, obtaining of its image is subjected to the influence of ambient lighting bigger, and needs the professional to carry out data analysis and identification indoor, and poor in timeliness can't satisfy concrete production practices requirement.
Plant disease diagnostic method based on expert system, it has proposed a kind of vegetables pest diagnosis expert system and has made up and method of servicing, the analysis mode of this method imitation plant protection expert pest diagnosis, adopted inference mechanism, founded the pest diagnosis expert system that comprises new case generation, case library formation, disease and pest symptom weight calculation, case retrieval and contents such as coupling, Man Machine Interface based on case.But the plant disease diagnostic method that is based on expert system exists diagnostic knowledge to obtain bottleneck, and coverage rate is limited, and knowledge rule upgrades difficult, and generalization ability is poor, deficiencies such as operational efficiency is low, system performance difference.
Based on the plant disease diagnostic method of fuzzy neural network, determine the degree of membership of corps diseases feature,, to infer the degree of membership of disease according to fuzzy technology by artificial neural network according to the degree of membership of disease symptom, show the tendentiousness that disease exists with this.
Artificial neural network has strong diagnostic knowledge learning ability, generalization ability and parallel processing capability in recent years, and is more noticeable, but still there are the following problems:
1, vegetable disease symptom performance is divided not comprehensive: research at present get more plant leaf, branches and tendrils, flower, really, the root system symptom, and part disease symptom to show be on seedling, show classical symptom in seedling stage as the cucumber samping off.
2, single utilization symptom importance weight (or degree of membership) makes up input vector, and it is not obvious to make in the sample that the diagnosis rule embodies, limited to the sick differentiation recognition capability of planting of part.As macrosporium leaf spot of red pepper and capsicum black spot, two kinds of diseases coexist and show classical symptom on the fruit, and other positions are asymptomatic, and at this moment both weights are identical, occur the corresponding different disease of identical input vector, neural network model easily and make up problem such as make mistakes.
3, classic method (additional momentum and adaptive learning rate method) is carried out the neural network performance optimization, is that an initial point from solution space begins to obtain the iterative search procedures of optimum solution.The search information that this single search point provided is few, and therefore, search efficiency is not high in the process of intention acquisition globally optimal solution.In addition, what these algorithms often used is the determinacy searching method, promptly a search point is put to another search and is had definite transfer method and transfer relationship, this determinacy often to make search procedure sink into locally optimal solution, even also might cause forever and can't reach optimum solution.
Summary of the invention
(1) technical matters that will solve
At above-mentioned defective, the technical problem to be solved in the present invention is how to improve the accuracy and the Fault-Tolerant Reasoning ability of vegetable disease diagnostic model.
(2) technical scheme
For addressing the above problem, the invention provides a kind of vegetable disease diagnostic method, described method comprises:
A: the vegetable plant strain with the performance of outward appearance morbid state is set up sympotomatic set and symptom importance degree of membership, wherein said sympotomatic set comprises 6 symptom subclass, each symptom subclass all with 6 subclass of position collection: root, stem are climing, leaf, flower, fruit and seedling are corresponding respectively, and described symptom importance degree of membership is used for characterizing the significance level that symptom is discerned disease;
B: obtain symptom characteristic, make up input vector in conjunction with described symptom importance degree of membership;
C: on described input vector basis, make up based on the Model Neural of improving genetic algorithm, and train, obtain ripe fuzzy neural network;
D: input vegetable disease symptom in the fuzzy neural network of described maturation, draw output vector, determine diagnostic result.
Preferably, described step B specifically comprises:
B1: at common semantic space the natural language description of described symptom is converted into term description, obtains the sample definition value;
B2: calculate vector with symptom characteristic and symptom material information according to described sample definition value and described symptom importance degree of membership;
B3: described vector with symptom characteristic and symptom material information is carried out brief processing, obtain input vector.
Preferably, described fuzzy neural network is used to carry out the diagnosis of disease, comprises input layer, obscuring layer, hidden layer and output layer;
Wherein said input layer is used for described sample definition value is input to described obscuring layer, and each node is represented an input variable;
Described obscuring layer is used to calculate the membership function of symptom for disease importance, with described input variable obfuscation, and obtain the sample definition value of symptom according to the vectorial construction method of common semantic space, based on symptom degree of membership and symptom sample definition value, original sample is carried out vectorization, obtain the initial optimization weights by genetic algorithm, pass to described hidden layer neuron with sample vector;
Described hidden layer is used for based on described sample vector and described initial optimization weights, calculates the mapping of described input layer to described output layer;
Described output layer is used for determining suspicious disease according to the maximum of described output vector to first position that if described maximal value approaches 0, then pairing disease possibility is little; If described maximal value approaches 1, then pairing disease possibility is big.
For addressing the above problem, the present invention also provides a kind of vegetable disease diagnosis portable system, and described system comprises:
Mobile terminal system, application back-up system and data communication middleware;
Wherein said mobile terminal system comprises that portable terminal and vegetable disease diagnose mobile application software; Described application back-up system comprises network experience service system, back-stage management back-up system and server; Described data communication middleware is used to connect described mobile terminal system and described application back-up system.
Preferably, described back-stage management back-up system comprises: sample data maintenance unit, diagnostic model training unit and service management unit;
Wherein said sample data maintenance unit comprises that original diagnostic sample editor module, sample input and output vector make up module and sample update notification module;
Described original diagnostic sample editor module, be used at original diagnostic sample editing process, import corresponding symptom respectively according to vegetable species, disease title and the different site of pathological change selected, and call based on the term mapping block of common semantic space natural language disease symptom is unified to describe, obtain symptom and describe the term sequence, deposit the symptom characteristic tables of data in;
Described sample input and output vector makes up module, be used for obtaining the definition value that symptom is described term from described symptom characteristic tables of data, and by calling symptom importance membership function, utilize input vector to make up formula again and obtain described input vector, make up output vector by calling " getting 1 among the n " scale-of-two discrimination bit modular converter, the input and output vector is deposited in the training sample value table at last;
Described sample update notification module is used for issuing the information of described training sample value table change.
Preferably, described sample input and output vector structure module specifically comprises: symptom processing module, input vector make up module and output vector makes up module;
Described symptom processing module, be used for the vegetable plant strain with the performance of outward appearance morbid state is set up sympotomatic set and definite symptom importance degree of membership, wherein said sympotomatic set comprises 6 symptom subclass, each symptom subclass all with 6 subclass of position collection: root, stem are climing, leaf, flower, fruit and seedling are corresponding respectively, and described symptom importance degree of membership is used for characterizing the significance level that symptom is discerned disease;
Described input vector makes up module, is used to obtain symptom characteristic, makes up input vector in conjunction with described symptom importance degree of membership;
Described output vector makes up module, is used for the binary coding method structure output vector of described disease centralized procurement with " n gets 1 ".
Preferably, described input vector structure module specifically comprises: conversion module, computing module and brief module;
Described conversion module is used at common semantic space the natural language description of described symptom being converted into term description, obtains the sample definition value;
Described computing module is used for calculating the vector with symptom characteristic and symptom material information according to described sample definition value and symptom importance degree of membership;
Described brief module is used for described vector with symptom characteristic and symptom material information is carried out brief processing, obtains input vector.
Described fuzzy neural network module comprises input layer, obscuring layer, hidden layer and output layer;
Wherein said input layer is used for described sample definition value is input to described obscuring layer, and each node is represented an input variable;
Described obscuring layer is used to calculate the membership function of symptom for disease importance, with described input variable obfuscation, and obtain the sample definition value of symptom according to the vectorial construction method of common semantic space, based on symptom degree of membership and symptom sample definition value, original sample is carried out vectorization obtain sample vector, obtain the initial optimization weights by genetic algorithm, pass to the neuron of described hidden layer with described sample vector;
Described hidden layer is used for based on described sample vector and described initial optimization weights, calculates the mapping of described input layer to described output layer;
Described output layer is used for determining suspicious disease according to the maximum of output vector to unit value position that if described maximal value approaches 0, then pairing disease possibility is little; If described maximal value approaches 1, then pairing disease possibility is big.
Preferably, described network experience service system is used to provide instructions to the user, the online experience of function and upgrading, database resource to download and renewal.
(3) beneficial effect
The present invention proposes a kind of vegetable disease diagnostic method and portable system, utilization is set up the input vector that symptom characteristic and symptom importance are described in carrying based on the input vector construction method of common semantic space, solve that sample set is representative poor, the contradiction sample is many, diagnose not obvious etc. the problem of rule embodiment, and in conjunction with making up fuzzy neural network model based on improving genetic algorithm, accuracy rate is higher as a result to make Model Diagnosis, and fault-tolerant ability is stronger.
Description of drawings
Fig. 1 is the process flow diagram of a kind of vegetable disease diagnostic method of the embodiment of the invention one;
Fig. 2 is the particular flow sheet of step B in a kind of vegetable disease diagnostic method of the embodiment of the invention one;
Fig. 3 is that input vector makes up exemplary plot in a kind of vegetable disease diagnostic method of the embodiment of the invention one;
Fig. 4 is the fuzzy neural network model figure in a kind of vegetable disease diagnostic method of the embodiment of the invention one;
Fig. 5 is that output vector makes up exemplary plot in a kind of vegetable disease diagnostic method of the embodiment of the invention one;
Fig. 6 is the input vector example of diagnostic test in a kind of vegetable disease diagnostic method of the embodiment of the invention one;
Fig. 7 is the output vector example of diagnostic test in a kind of vegetable disease diagnostic method of the embodiment of the invention one;
Fig. 8 is the composition synoptic diagram of a kind of vegetable disease diagnosis portable system of the embodiment of the invention two;
Fig. 9 is the surface chart that a kind of vegetable disease diagnosis portable system of the embodiment of the invention two is used;
Figure 10 is the composition synoptic diagram that a kind of vegetable disease of the embodiment of the invention two is diagnosed back-stage management back-up system in the portable system;
Figure 11 is the composition synoptic diagram that a kind of vegetable disease of the embodiment of the invention two is diagnosed sample input and output vector module in the portable system;
Figure 12 is the connection diagram that the input vector in the embodiment of the invention two makes up module, output vector structure module and fuzzy neural network.
Embodiment
Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.Following examples are used to illustrate the present invention, but are not used for limiting the scope of the invention.
Embodiment one
A kind of vegetable disease diagnostic method is provided in the embodiment of the invention one, and steps flow chart specifically may further comprise the steps as shown in Figure 1:
Steps A: the vegetable plant strain with the performance of outward appearance morbid state is set up sympotomatic set and symptom importance degree of membership.
Wherein, divide, each position of plant is represented with a set, in the general Study, the site of pathological change of plant is divided into root, stem, leaf, flower, 5 parts really according to plant morbidity performance.Because the part vegetables promptly fell ill in seedling stage, can determine its disease reason rapidly by the symptom performance in seedling stage, as the tomato samping off.Therefore, be to improve the comprehensive and accuracy of diagnosis, with vegetable plant strain morbidity performance be divided into finally that root, stem are climing, leaf, flower, really, 6 parts of seedling, the position set representations of morbidity performance is as follows:
P={P
i|i=1,2,3,4,5,6} (1)
In the formula, Pi represents that root, stem are climing, leaf, flower, really, one of 6 parts of seedling.
Symptom is the synergistic result of the plant and the cause of disease, and it is the main foundation that people discern disease, description disease and name disease.The corresponding sympotomatic set S of site of pathological change comprises 6 symptom subclass S
i, corresponding of 6 symptom subclass, stem are climing, leaf, flower, fruit and 6 positions of seedling, and symptom subclass Si is each morbidity performance related indication description in position.
S={S
i|i=1,2,3,4,5,6} (2)
By plant protection literature survey, typical case analysis, and agricultural experts seek advice from audit, the disease symptom is described the knowledge of expressing in the urtext data divides, form, leaf climing, flower by disease title and root, stem, really, the symptom bivariate table formed of seedling 6 positions.
Because different symptoms is to the percentage contribution difference of disease screening, the important evidence of certain disease is determined in some evident characteristic symptom performances often.Usually with fuzzy natural language the significance level of symptom for disease identification described.In this embodiment, it is divided into classical symptom〉cardinal symptom〉general three levels of symptom, be not confined in other embodiments of the invention symptom type identification degree is divided into other level, the contribution that each level is judged for symptom is all different, for example: classical symptom is for judging that the disease plant suffers from the decisive maximum of certain disease, can directly judge the kind of disease according to this symptom, cardinal symptom is for judging that the decisive of this illness will be inferior to classical symptom, just can not judge the disease type for certain according to main illness, judge the decisive minimum of disease species according to general symptom, almost can't judge the disease type according to general symptom.Can be divided in other embodiments and be not equal to 3 level.
Three symptom type explanations in the present embodiment are as shown in table 1.
The explanation of table 1 symptom type
Symptom importance subordinate function is defined as fuzzy language value, and symptom importance degree of membership is used for characterizing the significance level of symptom to disease identification, determines that according to the expertise method degree of membership of different levels is as follows:
Wherein L is the degree of membership of symptom subclass Si, and A, B, C are three kinds of symptom types.
Step B: obtain symptom characteristic, make up input vector in conjunction with symptom importance degree of membership.
Concrete, the steps flow chart of step B specifically may further comprise the steps as shown in Figure 2:
Step B1: at common semantic space the natural language description of symptom is converted into term description, obtains the sample definition value.
In conventional method, directly utilize the urtext of diagnosis data, with symptom performance position is the unit, carry out assignment and make up input vector as the sample component, the input vector that obtains so not only exists the letter amount of carrying few, and vectorial mould is different in size, the diagnosis rule embodies problems such as not obvious, but also be easy to generate the error situation of the corresponding different diseases kind of identical sample value, and can not distinguish preferably the disease reason, this also certainly will have influence on the accuracy of diagnosis.
And in the present embodiment, the natural language sample of source book is mapped in the common semantic space, the relational language that utilizes the symptom illness is described the urtext of symptom data, utilize disease symptom term coding schedule, the symptom bivariate table is converted into botany term description bivariate table in the common semantic space, and the sample value of determining the natural language sample according to the definition value and the symptom importance degree of membership of term, thereby structure input vector, can effectively enrich the vector information bearing capacity, give full expression to the diagnosis rule.
Symptom subclass Si is each morbidity performance related indication description in position, the symptom characteristic that all comprises vegetable plant strain site of pathological change among each symptom subclass Si with the performance of outward appearance morbid state, wherein symptom characteristic comprises symptom and illness, and the sample definition value is exactly the vector of being made up of symptom term coding and illness sorting code number.
At first to carry out natural language symptom mapping based on common semantic space.Primitive nature language disease symptom data is unified to describe in common semantic space.According to botany knowledge, the ill performance of the outward appearance of infected plant can be divided into symptom and symptom two big classes.Wherein symptom is exactly the state of seeing in sick portion, common as variable color, necrosis, wilting, rot and deformity etc., the spot of concrete manifestation such as brown, transparent striped, branches and leaves are wilted or tumour etc.; Symptom is meant the fructification of the pathogen that occurs in sick portion, the bacterial ooze of the mycelium, sclerotium, spore device of fungi, black powder, white powder, rust shape thing, mustiness thing, bacterium etc. is for example arranged.
According to disease symptom table, the natural language description of symptom is converted into term description.As climing natural language symptom of capsicum droop stem={ water soaking mode rots, and back complete stool is withered, sick portion white mustiness thing }, after the semantic space mapping, S
2=web rot, and withered, the mustiness thing }, its sample definition value D (S
2) be 0,0,2,2,0,0,1}.
The disease symptom term and the coding schedule of common semantic space are as shown in table 2:
Table 2 symptom term and coding schedule
Step B2: calculate vector with symptom characteristic and symptom material information according to sample definition value and symptom importance degree of membership.
Wherein, symptom importance degree of membership is used for representing the significance level of symptom for disease identification, and represents with the symptom importance degree of membership that varies in size for the symptom of different significance levels.
Comprehensive sample definition value D (S
i) and symptom importance degree of membership L (S
i), form vector with symptom characteristic and symptom material information, be input vector, be expressed as:
X
i={D(S
1)*L(S
1),D(S
2)*L(S
2),……,D(S
i)*L(S
i)}
Wherein, D (S
i) be S
iThe sample definition value of symptom, L (Si) is the importance degree of membership of Si symptom, i is the input vector dimension.
Step B3: the vector with symptom characteristic and symptom material information is carried out brief processing, obtain input vector.
Handle counting yield in order to improve the input vector matrix at the later stage neural network model, remove in the matrix and to be 0 no information row with train value,, improve operational efficiency to reduce matrix dimensionality.The final input vector matrix of brief formation.Learn diagnosis with the input vector data of 19 kinds of diseases of tomato in the present embodiment, input vector as shown in Figure 3.
Step C: on the input vector basis that step S1 obtains, make up based on the fuzzy neural network of improving genetic algorithm, and train, obtain ripe fuzzy neural network.
Obtain ripe fuzzy neural network after waiting to train based on the fuzzy neural network of improving genetic algorithm, be used to carry out the vegetable disease diagnosis, it comprises input layer, obscuring layer, hidden layer and output layer, adopt fuzzy BP neural network to make up, fuzzy system is connected with the mode of neural network according to series connection, with fuzzy system original knowledge is carried out pre-treatment, carry out disease screening with neural network.
Fuzzy neural network model as shown in Figure 4, input layer wherein is used for the sample definition value is input to obscuring layer, each node is represented an input variable;
Obscuring layer is used to calculate the membership function of symptom for disease importance, with the input variable obfuscation, and obtain the sample definition value of symptom according to the vectorial construction method of common semantic space, based on symptom degree of membership and symptom sample definition value, original sample is carried out vectorization obtain sample vector, obtain the initial optimization weights by improving genetic algorithm, pass to the neuron of hidden layer with sample vector.
The improved method of genetic algorithm is: for the genetic algorithm of the real coding of routine, the span of interaction coefficent is [0,1], when optimizing the weights of neural network model with it, can cause the chromosome individuality in the search volume, to shrink or disperse, therefore the optimization ability reduces, and is unfavorable for global optimization.For amount of variability, if amount of variability is too little, search area is too narrow, and evolutionary rate is slow; If amount of variability is too big, search near local vibration impact point, can't reach optimization aim.To this, in the present embodiment genetic algorithm of optimizing the neural network model weights is improved, its basic ideas are: for being in the initial evenly parent chromosome individuality of distribution, after interlace operation, increase the dispersion degree of population, make it produce new search space, allow the offspring individual that it produced in search space, also evenly distribute simultaneously, in the search volume, neither shrink and also do not disperse, and guarantee that its individuality with defect mode is not destroyed fully by interlace operation institute, thereby improve optimization ability in the overall situation.For mutation operator, when evolutionary process is smooth, aberration rate is maintained a lower level; When evolutionary process is not smooth, by improving aberration rate, to enlarge the hunting zone.When evolutionary generation reaches a certain threshold value of setting, think that colony has been absorbed in earliness, make colony break away from earliness by impressed variation, its step is as follows:
(1) carries out the chromogene coding according to the structure of neural network model and the problem of finding the solution.
(2) generate initial equally distributed chromosome population at random.
(3) according to the problem of being found the solution, calculate the fitness value of individual in population, with training sample the neural network of individual in population representative is trained, calculate the study error of each individuality, thereby determine fitness value.The study error E adopts following formula to calculate:
N is a number of training, and q is the output unit number,
Expression is poor with the actual output of j output unit of k sample training and desired output.Fitness function fs=1/E, it guarantees that error is more little, fitness value is big more.
(4) individual in population is carried out genetic manipulation, comprise selection, intersect and variation.
1. select: carry out selection operation with the roulette method.
2. intersect:, be interval V=[xmin, xmax if x1, x2 are the parent individuality] go up equally distributed random number, then intersect the back interval (xmin, xmax) go up equally distributed random number z1, z2 by formula decision down:
y1=ax1+(1-a)x2，y2=ax2+(1-a)x1
z1=MOD(y1，V)，z2=MOD(y2，V)
Z1, z2 are the offspring individuals after intersecting, and wherein a is an integer, and MOD is the modulo operation symbol.
3. variation: aberration rate Pm is defined by following formula:
Pm=0.001+NG*cof
In the following formula, Pm represents the aberration rate of current algebraically; NG is the algebraically of not evolved continuously till since evolving last time; Cof is the coefficient of decision chromosome impressed variation (100% variation) threshold value.Amount of variability var is defined as:
var=rand*w*dyna
In the formula, rand is the random number between [0,1] that produces at random; W is a fixed value in the span of power; Dyna is the dynamic parameter of decision amount of variability var, initial dyna=1.0; If counter〉nochange, then dyna=dyna0.1 and counter=0; Wherein, counter is a counter, and statistics is extremely current on behalf of ending the algebraically of not evolving continuously since evolving last time; Nochange is a constant, is to judge the threshold value that whether changes dyna.
(5) calculating once more of fitness value.
(6), iterate and stop the optimum solution of output problem if satisfy the criterion that stops search; Otherwise, turn to step (4).
Simultaneously, for guaranteeing that the individuality of good pattern is not destroyed by genetic manipulation, adopt the optimum individual retention strategy, the individuality of promptly current fitness value maximum directly entails the next generation without genetic manipulation.Improved genetic algorithm has guaranteed to optimize the stability and the convergence of neural network model, makes that also the genetic algorithm after improving has higher optimization efficient all the time in the process of optimizing neural network model.Simultaneously, utilize amended genetic algorithm to calculate the speed that can improve calculating, thereby improve model calculation efficient.
Hidden layer is used for based on the sample vector of obscuring layer and initial optimization weights, calculates the mapping of input layer to output layer.The number of hidden nodes determines that method is as follows:
Wherein: l is the hidden neuron number, and n is the input layer number, and m is the output layer neuron number, and a is the constant between the value 0 to 10.
Output layer is used for determining suspicious disease according to the maximum of output vector to the position of unit that if maximal value approaches 0, then pairing disease possibility is little; If maximal value approaches 1, then pairing disease possibility is big.
Step D: input vegetable disease symptom in the fuzzy neural network of maturation, draw output vector, determine diagnostic result.
The sample output vector that wherein is used for carrying out the fuzzy neural network training is that the binary coding method of disease centralized procurement with " n gets 1 " made up, as shown in Figure 5, concrete construction method is to compile various infectivities and the noninfectious disease that causes that vegetable plant strain causes a disease, set up the disease collection, be expressed as follows: Y={Y
1, Y
2..., Yn}
According to the kind n of disease, determine that length is the binary coding of n.Only have 1 to be 1 in every group coding, all the other n-1 positions are 0, represent a certain disease.In this coding method, 1 diverse location is corresponding one by one with the different diseases kind.
Actual vegetable disease symptom is converted into input vector, is input in the ripe fuzzy neural network model of training, obtain output vector.Determine method (being that the output vector maximum is to first position sign disease species) according to disease, output vector is carried out decipher, draw diagnostic result.
Also to carry out diagnostic model training, model accuracy test and the test of model fault-tolerance for the fuzzy neural network model of setting up in the present embodiment.
Diagnostic model training: based on fuzzy neural network model, in the C# programmed environment, call the M file exploitation neural metwork training module of Matlab, make the maturing of vegetable disease diagnostic model also can drop into application.Concrete grammar is: determine cardinal rule input hidden layer node number according to hidden layer neuron, according to disease species output layer node number is set, call the examining training sample of corresponding crop, button click carries out network training.System shows frequency of training, final error, error changing trend diagram information in real time.When training error reaches target error, this network training maturation is so far finished training.
Model accuracy test: the general neural network diagnostic model that directly utilizes the symptom weights as input vector in this research and the conventional method is carried out accuracy relatively.Test data comprises two classes, i.e. the data of laboratory field data data generation, and user concerning farmers carries out the data that the symptom selection operation generates according to practical condition.Through plant protection expert checking, obtain test result mean value and see as shown in table 3.
Table 3 accuracy experiment test result
In the statistical result showed, indoor and outdoor test, all increase aspect the accuracy based on the general neural network of diagnostic method of fuzzy neural network, the thinking scheme that this research is described is effective.Wherein, the field data data test result that the laboratory utilized is better than peasant household's practical application.Its reason is, the employed field data data in laboratory is near the diagnostic knowledge in the documents and materials, and has fault-tolerance preferably based on the vegetable disease model of fuzzy neural network, so accuracy of diagnosis is higher.The symptom performance that peasant household of outside basic unit then sees according to oneself is fully aborning carried out selection operation and is formed test data, has reflected the practical situations of model more really.Because exist a plurality of diseases to be mingled with the complex situations of performance simultaneously in the actual production, this certain degree has influenced accuracy of diagnosis.Therefore also explanation in effort aspect this, can further improve the practicality of model.
Model fault-tolerance test: in actual application, the disease symptom that the user provides can't be in full accord with sample, the possibility maximum that the disease classical symptom is selected, but part cardinal symptom and general symptom exist A that (provide symptom and sample symptom inconsistent) be provided, B multiselect (providing symptom) more than the sample symptom, C is provided by (providing symptom to be less than the sample symptom) less, A+B multiselect and falsely dropping, the situation that A+C selects less and falsely drops, choose 5 groups of representative test datas of user in view of the above, present embodiment is example with the canker of tomato, the input vector of diagnostic test as shown in Figure 6, come the fault-tolerance of testing model, output result such as Fig. 7.
In 19 latitude output vectors shown in Fig. 7, what value was maximum has identified disease species to first position.What the 17th was in the sample output vector is maximal value to unit, shows that then this output result is a canker of tomato.In 5 groups of representative user test data, output vector to first maximal value all the time at the 17th, illustrated that diagnostic model has stronger fault-tolerant ability, this also is this method key high than the additive method diagnostic accuracy.Simultaneously, when the user " falsely drops ", " multiselect ", and when " multiselect+falsely drop ", the 17th of output vector is respectively 0.9999,0.9876,0.9216 to unit's value, near the sample analogue value 1; As user " few choosing ", and when " selecting less+falsely drop ", the 17th of output vector is respectively 0.7786,0.5946 to unit's value, than other group test datas, away from the sample analogue value 1, illustrate that the user provides disease symptom information many more, the possibility of correctly diagnosing is big more.
Pass through said method, utilization is set up the input vector that symptom characteristic and symptom importance degree of membership are described in carrying based on the input vector construction method of common semantic space, solve problems such as sample set is representative poor, the contradiction sample is many, the embodiment of diagnosis rule is not obvious, and in conjunction with the fuzzy neural network model that utilizes based on improved genetic algorithm, make that vegetable disease diagnostic result accuracy rate is higher, the Fault-Tolerant Reasoning ability is stronger.
Embodiment two
For achieving the above object, a kind of vegetable disease diagnosis portable system also is provided in the embodiments of the invention two, form synoptic diagram as shown in Figure 8, specifically comprise:
Mobile terminal system 810, application back-up system 820 and data communication middleware 830.
Wherein mobile terminal system 810 comprises that portable terminal 811 and vegetable disease diagnose mobile application software 812.Use back-up system 820 and comprise network experience service system 821, back-stage management back-up system 822 and server 823.
Data communication middleware 830 is used to connect mobile terminal system 810 and uses back-up system 820.In order to realize the data access of portable system and PC end, based on ASP.NET+ADO.NET development data communication middleware.It between PC and palm PC, with functions such as the access of realization virtual buffer, format conversion, decompress(ion)s, thereby carries out resource sharing between the two in logic.
Wherein portable terminal 811 can be palm PC, can also be intelligent mobile phone terminal.The portable terminal of selecting for use in the present embodiment is a palm PC, and system interface as shown in Figure 9.
Aspect software, this application system breaks through the form of General System with literal guiding and user's input, adopt the mode of picture expression and literal aid illustration, make the user can obtain diagnostic result by simple picture browsing and clicking operation, complicated computing is given the backstage and is finished, and is more suitable for the user group based on the peasant like this.In the concrete diagnostic procedure, the user at first selects vegetable species, shows under the tab of position six diseases one by one then, according to the picture and the text description at this position, selects, and submits to then to get final product.System diagnoses by the model that this kind of routine call vegetables have trained automatically, and provides the kind and the possibility size thereof of disease, simultaneously, calls the Prevention Technique measure of this kind disease in the database, forms vegetable disease diagnosis electronic prescription and signs.
At hardware aspect, select that Costco Wholesale is low, display quality good for use, screen pure and fresh generous, possess secondary application and development support function, satisfy parameter request with the palm PC that ensures application system stable operation hardware carrier as the vegetable disease diagnostic system.Major parameter requires as follows:
1. operating system: Windows CE;
2. processor host frequency: more than the 400MHZ;
3. input mode: hand-written and keyboard input
4. memory capacity: more than the 1GB, carry out the storage of diagnostic knowledge base and relevant disease picture resource.
5. data-interface USB
6. network schemer: support 3G and WIFI;
7. screen size: 7 cun.
Network is experienced service system 821 and is used to provide instructions to the user, the online experience of function and upgrading, database resource to download and renewal.
For helping the user to understand better and use vegetable disease diagnosis portable system, supporting on the internet exploitation corresponding network is experienced and service prefecture system, and following adequate and systematic service is provided:
1. palm PC mobile end system function declaration and download: to the download and the using method of system, carry out introduction of substep picture and text and explanation, mobile terminal operational system download link is provided simultaneously, use with the guiding user installation.
2. online experience of function and upgrading: the selected part vegetable species, network edition inline diagnosis link is provided, make the user that the service that is provided can more be provided in depth.While issuing function upgrade notification is so that the user can be applied to latest edition immediately.
3. online download of database resource and renewal: constantly perfect along with diagnostic knowledge base, diagnosable vegetable species is more and more, and experiencing service area at network will provide more comprehensively rich in natural resources to download and upgrade.
The composition synoptic diagram of back-stage management back-up system 822 specifically comprises as shown in figure 10: sample data maintenance unit 8221, diagnostic model training unit 8222 and service management unit 8223.For better grasping user's operating position, obtain field feedback, improve the practicality of system, background management system is carried out authentication registration to the user, and carries out download statistics, in-service evaluation analysis.
Wherein, sample data maintenance unit 8221 comprises: original diagnostic sample editor module 82211, sample input and output vector make up module 82212 and sample update notification module 82213.
Original diagnostic sample editor module 82211, be used at original diagnostic sample editing process, import corresponding symptom respectively according to vegetable species, disease title and the different site of pathological change selected, and call based on the term mapping block of common semantic space natural language disease symptom is unified to describe, obtain symptom and describe the term sequence, deposit the symptom characteristic tables of data in;
Sample input and output vector makes up module 82212, be used for obtaining the definition value that symptom is described term from the symptom characteristic tables of data, and by calling symptom importance membership function and symptom characteristic coding, utilize input vector to make up formula and obtain input vector, make up output vector by calling " getting 1 among the n " scale-of-two discrimination bit modular converter, the input and output vector is deposited in the training sample value table at last;
Sample update notification module 82213 is used for issuing the information that training sample value table changes, as modification, interpolation etc.
The composition synoptic diagram of sample input and output vector structure module 82212 specifically comprises as shown in figure 11: symptom processing module 822121, input vector make up module 822122 and output vector structure module 822123.
Symptom processing module 822121 is used for the vegetable plant strain with the performance of outward appearance morbid state is set up sympotomatic set and definite symptom importance degree of membership.
Input vector makes up module 822122, is used to obtain symptom characteristic, makes up input vector in conjunction with symptom importance degree of membership.
Wherein, input vector structure module 822122 specifically comprises: conversion module 8221221, computing module 8221222 and brief module 8221223.
Conversion module 8221221 is used at common semantic space the natural language description of symptom being converted into term description, obtains the sample definition value.
Computing module 8221222 is used for calculating the vector with symptom characteristic and symptom material information according to sample definition value and symptom importance degree of membership.
Brief module 8221223 is used for the vector with symptom characteristic and symptom material information is carried out brief processing, obtains input vector.
Fuzzy neural network is used to carry out the diagnosis of disease, comprises input layer, obscuring layer, hidden layer and output layer.
Because symptom comprises symptom and illness, corresponding input vector, the disease set pair is answered output vector, after input vector and output vector make up and finish, make up fuzzy neural network again, so the connection diagram of input vector structure module 822122 and output vector structure module 822123 and fuzzy neural network as shown in figure 12.
Wherein input layer is used for the sample definition value is input to obscuring layer, and each node is represented an input variable.
Obscuring layer is used to calculate the membership function of symptom for disease importance, with the input variable obfuscation, and obtain the sample definition value of symptom according to the vectorial construction method of common semantic space, based on symptom degree of membership and symptom sample definition value, original sample is carried out vectorization obtain sample vector, obtain the initial optimization weights by genetic algorithm, pass to the neuron of hidden layer with sample vector.
Hidden layer is used for based on obscuring layer sample vector and initial optimization weights, calculates the mapping of input layer to output layer.
Output layer is used for determining suspicious disease according to the maximum of output vector to the position of unit, and is little if maximal value approaches 0 pairing disease possibility; If it is big that maximal value approaches 1 pairing disease possibility.
Output vector makes up module 822123, is used for the fuzzy neural network input vegetable disease symptom in trained maturation, draws output vector, to the binary coding method structure output vector of disease centralized procurement with " getting 1 among the n ", and definite diagnostic result.
By using said system, utilization is set up the input vector that symptom characteristic and symptom importance are described in carrying based on the input vector construction method of common semantic space, solve problems such as sample set is representative poor, the contradiction sample is many, the embodiment of diagnosis rule is not obvious, and in conjunction with making up fuzzy neural network model based on improving genetic algorithm, make that vegetable disease diagnostic result accuracy rate is higher, the Fault-Tolerant Reasoning ability is stronger.
Above embodiment only is used to illustrate the present invention; and be not limitation of the present invention; the those of ordinary skill in relevant technologies field; under the situation that does not break away from the spirit and scope of the present invention; can also make various variations and modification; therefore all technical schemes that are equal to also belong to category of the present invention, and scope of patent protection of the present invention should be defined by the claims.
Claims (10)
1. a vegetable disease diagnostic method is characterized in that, described method specifically comprises:
A: the vegetable plant strain with the performance of outward appearance morbid state is set up sympotomatic set and symptom importance degree of membership, wherein said sympotomatic set comprises 6 symptom subclass, each symptom subclass is all climing with position collection root, stem, leaf, flower, fruit and seedling are corresponding respectively, and described symptom importance degree of membership is used for characterizing the significance level that symptom is discerned disease;
B: obtain symptom characteristic, make up input vector in conjunction with described symptom importance degree of membership;
C: on described input vector basis, make up based on the Model Neural of improving genetic algorithm, and train, obtain ripe fuzzy neural network;
D: input vegetable disease symptom in the fuzzy neural network of described maturation, draw output vector, determine diagnostic result.
2. the method for claim 1 is characterized in that, described step B specifically comprises:
B1: at common semantic space the natural language description of described symptom is converted into term description, obtains the sample definition value;
B2: calculate vector with symptom characteristic and symptom material information according to described sample definition value and described symptom importance degree of membership;
B3: described vector with symptom characteristic and symptom material information is carried out brief processing, obtain input vector.
3. the method for claim 1, it is characterized in that, comprise described symptom characteristic in the described symptom subclass with vegetable plant strain site of pathological change of outward appearance morbid state performance, described symptom characteristic comprises symptom and illness, the vector that described sample definition value is made up of described symptom description coding and illness sorting code number.
4. the method for claim 1 is characterized in that, described fuzzy neural network is used to carry out the diagnosis of disease, comprises input layer, obscuring layer, hidden layer and output layer;
Wherein said input layer is used for described sample definition value is input to described obscuring layer, and each node is represented an input variable;
Described obscuring layer is used to calculate the membership function of symptom for disease importance, with described input variable obfuscation, and obtain the sample definition value of symptom according to the vectorial construction method of common semantic space, based on symptom degree of membership and symptom sample definition value, original sample is carried out vectorization, obtain the initial optimization weights by genetic algorithm, pass to described hidden layer neuron with sample vector;
Described hidden layer is used for based on described sample vector and described initial optimization weights, calculates the mapping of described input layer to described output layer;
Described output layer is used for determining suspicious disease according to the maximum of described output vector to first position that if described maximal value approaches 0, then pairing disease possibility is little; If described maximal value approaches 1, then pairing disease possibility is big.
5. a vegetable disease diagnosis portable system is characterized in that described system specifically comprises: mobile terminal system, application back-up system and data communication middleware;
Wherein said mobile terminal system comprises that portable terminal and vegetable disease diagnose mobile application software; Described application back-up system comprises network experience service system, back-stage management back-up system and server; Described data communication middleware is used to connect described mobile terminal system and described application back-up system.
6. portable system as claimed in claim 5 is characterized in that, described back-stage management back-up system comprises: sample data maintenance unit, diagnostic model training unit and service management unit;
Wherein said sample data maintenance unit comprises that original diagnostic sample editor module, sample input and output vector make up module and sample update notification module;
Described original diagnostic sample editor module, be used at original diagnostic sample editing process, import corresponding symptom respectively according to vegetable species, disease title and the different site of pathological change selected, and call based on the term mapping block of common semantic space natural language disease symptom is unified to describe, obtain symptom and describe the term sequence, deposit the symptom characteristic tables of data in;
Described sample input and output vector makes up module, be used for obtaining the definition value that symptom is described term from described symptom characteristic tables of data, and by calling symptom importance membership function, utilize input vector to make up formula again and obtain described input vector, make up output vector by calling " getting 1 among the n " scale-of-two discrimination bit modular converter, the input and output vector is deposited in the training sample value table at last;
Described sample update notification module is used for issuing the information of described training sample value table change.
7. portable system as claimed in claim 6 is characterized in that, described sample input and output vector makes up module and specifically comprises: symptom processing module, input vector make up module and output vector makes up module;
Described symptom processing module, be used for the vegetable plant strain with the performance of outward appearance morbid state is set up sympotomatic set and definite symptom importance degree of membership, wherein said sympotomatic set comprises 6 symptom subclass, each symptom subclass all with 6 subclass of position collection: root, stem are climing, leaf, flower, fruit and seedling are corresponding respectively, and described symptom importance degree of membership is used for characterizing the significance level that symptom is discerned disease;
Described input vector makes up module, is used to obtain symptom characteristic, makes up input vector in conjunction with described symptom importance degree of membership;
Described output vector makes up module, is used for the binary coding method structure output vector of described disease centralized procurement with " n gets 1 ".
8. portable system as claimed in claim 7 is characterized in that, described input vector makes up module and specifically comprises: conversion module, computing module and brief module;
Described conversion module is used at common semantic space the natural language description of described symptom being converted into term description, obtains the sample definition value;
Described computing module is used for calculating the vector with symptom characteristic and symptom material information according to described sample definition value and symptom importance degree of membership;
Described brief module is used for described vector with symptom characteristic and symptom material information is carried out brief processing, obtains input vector.
9. portable system as claimed in claim 7 is characterized in that, described fuzzy neural network based on genetic algorithm comprises input layer, obscuring layer, hidden layer and output layer;
Wherein said input layer is used for described sample definition value is input to described obscuring layer, and each node is represented an input variable;
Described obscuring layer is used to calculate the membership function of symptom for disease importance, with described input variable obfuscation, and obtain the sample definition value of symptom according to the vectorial construction method of common semantic space, based on symptom degree of membership and symptom sample definition value, original sample is carried out vectorization obtain sample vector, obtain the initial optimization weights by genetic algorithm, pass to the neuron of described hidden layer with described sample vector;
Described hidden layer is used for based on described sample vector and described initial optimization weights, calculates the mapping of described input layer to described output layer;
Described output layer is used for determining suspicious disease according to the maximum of output vector to unit value position that if described maximal value approaches 0, then pairing disease possibility is little; If described maximal value approaches 1, then pairing disease possibility is big.
10. portable system as claimed in claim 5 is characterized in that, described network is experienced service system and is used to provide instructions to the user, the online experience of function and upgrading, database resource to download and renewal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310084471.0A CN103218748B (en) | 2013-03-15 | 2013-03-15 | A kind of vegetable disease diagnostic method and portable system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310084471.0A CN103218748B (en) | 2013-03-15 | 2013-03-15 | A kind of vegetable disease diagnostic method and portable system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103218748A true CN103218748A (en) | 2013-07-24 |
| CN103218748B CN103218748B (en) | 2016-08-31 |
Family
ID=48816505
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310084471.0A Active CN103218748B (en) | 2013-03-15 | 2013-03-15 | A kind of vegetable disease diagnostic method and portable system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103218748B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104636981A (en) * | 2013-11-08 | 2015-05-20 | 财团法人资讯工业策进会 | Treatment mode suggestion system and method for plant symptoms |
| CN108229413A (en) * | 2018-01-16 | 2018-06-29 | 宁夏智启连山科技有限公司 | Pest and disease damage kind identification method and device |
| CN109344738A (en) * | 2018-09-12 | 2019-02-15 | 杭州睿琪软件有限公司 | The recognition methods of crop diseases and pest crop smothering and device |
| CN109738438A (en) * | 2018-12-29 | 2019-05-10 | 扬州大学 | A kind of Wheat Leaves Senescence degree method for fast measuring |
| CN110598841A (en) * | 2018-06-13 | 2019-12-20 | 南京大学 | Flower disease analysis method based on multi-input convolutional neural network |
| CN110837926A (en) * | 2019-11-04 | 2020-02-25 | 四川省烟草公司广元市公司 | Tobacco main pest and disease damage prediction method based on big data |
| CN112052905A (en) * | 2020-09-11 | 2020-12-08 | 重庆科技学院 | Method for extracting multi-operation fatigue features of driver based on recurrent neural network |
| CN113449893A (en) * | 2020-03-25 | 2021-09-28 | 中移(成都)信息通信科技有限公司 | Insect pest prediction model training method, insect pest prediction method and insect pest prediction device |
| CN115601585A (en) * | 2022-09-15 | 2023-01-13 | 中科聚信信息技术（北京）有限公司（Cn） | Agricultural pest and disease diagnosis method and device based on picture analysis |
| CN116636815A (en) * | 2023-06-06 | 2023-08-25 | 中国人民解放军海军特色医学中心 | Electroencephalogram signal-based sleeping quality assessment method and system for underwater operators |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101881726A (en) * | 2010-06-18 | 2010-11-10 | 北京农业智能装备技术研究中心 | Nondestructive detection method for comprehensive character living bodies of plant seedlings |
| CN101894320A (en) * | 2010-06-30 | 2010-11-24 | 兰雨晴 | Agricultural activity support system based on mobile communication network and application method thereof |
-
2013
- 2013-03-15 CN CN201310084471.0A patent/CN103218748B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101881726A (en) * | 2010-06-18 | 2010-11-10 | 北京农业智能装备技术研究中心 | Nondestructive detection method for comprehensive character living bodies of plant seedlings |
| CN101894320A (en) * | 2010-06-30 | 2010-11-24 | 兰雨晴 | Agricultural activity support system based on mobile communication network and application method thereof |
Non-Patent Citations (1)
| Title |
|---|
| 夏于等: "基于物联网的小麦苗情诊断管理系统设计与实现", 《农业工程学报》, vol. 29, no. 5, 1 March 2013 (2013-03-01) * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104636981A (en) * | 2013-11-08 | 2015-05-20 | 财团法人资讯工业策进会 | Treatment mode suggestion system and method for plant symptoms |
| CN108229413A (en) * | 2018-01-16 | 2018-06-29 | 宁夏智启连山科技有限公司 | Pest and disease damage kind identification method and device |
| CN110598841A (en) * | 2018-06-13 | 2019-12-20 | 南京大学 | Flower disease analysis method based on multi-input convolutional neural network |
| CN109344738A (en) * | 2018-09-12 | 2019-02-15 | 杭州睿琪软件有限公司 | The recognition methods of crop diseases and pest crop smothering and device |
| CN109738438A (en) * | 2018-12-29 | 2019-05-10 | 扬州大学 | A kind of Wheat Leaves Senescence degree method for fast measuring |
| CN110837926B (en) * | 2019-11-04 | 2022-08-12 | 四川省烟草公司广元市公司 | Tobacco main pest and disease damage prediction method based on big data |
| CN110837926A (en) * | 2019-11-04 | 2020-02-25 | 四川省烟草公司广元市公司 | Tobacco main pest and disease damage prediction method based on big data |
| CN113449893A (en) * | 2020-03-25 | 2021-09-28 | 中移(成都)信息通信科技有限公司 | Insect pest prediction model training method, insect pest prediction method and insect pest prediction device |
| CN112052905A (en) * | 2020-09-11 | 2020-12-08 | 重庆科技学院 | Method for extracting multi-operation fatigue features of driver based on recurrent neural network |
| CN112052905B (en) * | 2020-09-11 | 2023-02-03 | 重庆科技学院 | Method for extracting multi-operation fatigue features of driver based on recurrent neural network |
| CN115601585A (en) * | 2022-09-15 | 2023-01-13 | 中科聚信信息技术（北京）有限公司（Cn） | Agricultural pest and disease diagnosis method and device based on picture analysis |
| CN116636815A (en) * | 2023-06-06 | 2023-08-25 | 中国人民解放军海军特色医学中心 | Electroencephalogram signal-based sleeping quality assessment method and system for underwater operators |
| CN116636815B (en) * | 2023-06-06 | 2024-03-01 | 中国人民解放军海军特色医学中心 | Electroencephalogram signal-based sleeping quality assessment method and system for underwater operators |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103218748B (en) | 2016-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103218748A (en) | Diagnostic method of vegetable diseases and portable system | |
| Wu et al. | Evolving RBF neural networks for rainfall prediction using hybrid particle swarm optimization and genetic algorithm | |
| Fan et al. | Development of prediction models for next-day building energy consumption and peak power demand using data mining techniques | |
| Overmars et al. | Comparison of a deductive and an inductive approach to specify land suitability in a spatially explicit land use model | |
| Sohn et al. | Decision tree based on data envelopment analysis for effective technology commercialization | |
| Walsh et al. | Complexity theory, spatial simulation models, and land use dynamics in the Northern Ecuadorian Amazon | |
| Wainwright et al. | Mind, the gap in landscape‐evolution modelling | |
| Hakli | BinEHO: a new binary variant based on elephant herding optimization algorithm | |
| Caplat et al. | Symmetric competition causes population oscillations in an individual-based model of forest dynamics | |
| Xiong et al. | Multiple peer effects in the diffusion of innovations on social networks: a simulation study | |
| CN107046557A (en) | The intelligent medical calling inquiry system that dynamic Skyline is inquired about under mobile cloud computing environment | |
| Guo et al. | iRedistrict: Geovisual analytics for redistricting optimization | |
| Lekar et al. | Construction of bayesian networks in public administration of the economy | |
| CN110263183A (en) | A kind of smart home DSS and decision support method | |
| Balbi et al. | Reviewing agent-based modelling of socio-ecosystems: a methodology for the analysis of climate change adaptation and sustainability | |
| Walisadeera et al. | Designing a Farmer Centred Ontology for Social Life Network. | |
| Kemp-Benedict et al. | Using matching methods to link social and physical analyses for sustainability planning | |
| Dhabal et al. | Crop recommendation system with cloud computing | |
| CN106228235A (en) | A kind of land utilization space Optimal Configuration Method taking general layout and PROCESS COUPLING effect into account | |
| Wang et al. | RETRACTED ARTICLE: Intrusion detection and performance simulation based on improved sequential pattern mining algorithm | |
| Hollmig et al. | An evaluation of combinations of lossy compression and change-detection approaches for time-series data | |
| Helfer et al. | Ubiquitous computing in precision agriculture: A systematic review | |
| Lu et al. | Image classification and identification for rice leaf diseases based on improved WOACW_SimpleNet | |
| Gao | Design and Implementation of 3D Animation Data Processing Development Platform Based on Artificial Intelligence | |
| Keskin et al. | FSOLAP: A fuzzy logic-based spatial OLAP framework for effective predictive analytics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |