公开(公告)号：EP1193645B1

公开(公告)日：2007-12-12

申请号：EP00830647.4

申请日：2000-10-02

Applicant: STMicroelectronics S.r.l.

Inventor： Pappalardo, Francesco ,  Giacalone, Biagio ,  Palano, Carmelo Marcello ,  Luzzi, Claudio ,  Grande, Francesca

IPC: G06N7/04 ,  G06F1/035

CPC classification number: G06F1/035 ,  G06N7/04

公开(公告)号：EP1124199A1

公开(公告)日：2001-08-16

申请号：EP00830082.4

申请日：2000-02-08

Applicant: STMicroelectronics S.r.l.

Inventor： Pappalardo, Francesco ,  Giacalone, Biagio ,  Palano, Carmelo

IPC: G06N7/04

CPC classification number: G06N7/04

Abstract: For encoded membership functions used to identify the atomic conditions defining the antecedents of fuzzy inferences, and also for the determination of the operands of the said antecedents, corresponding stores (6, 9) are configured for the storage of the already available values of these encoded membership functions and of the said operands. At the time of identification of a new value for the said quantities, a check is made (5, 7) to see whether this value is already present in the corresponding store (6, 9). If the outcome of this check is positive, in the case of encoded membership functions, the mechanism by which the encoded fuzzy inferences point to these functions is changed, so that the pointers of the encoded fuzzy inferences are redirected towards the membership functions which are already stored. In the case of the operands of the antecedents, the check of the corresponding back-up store (9) is carried out preferably on the basis of the corresponding calculation values, the calculation of a new operand being disabled when it is found that the corresponding calculation parameters are already present in the corresponding back-up store (9). The aforesaid store (9) is preferably configured as a stack to which new calculated operand values are written at the uppermost position, with the possibility of making values identified as already present shift back to this uppermost position.

Abstract translation: 对于用于识别限定模糊推理的前提的原子条件的编码的隶属函数，以及用于确定所述前提的操作数，对应的存储（6,9）被配置用于存储这些编码的已经可用的值 隶属函数和所述操作数。 在识别所述数量的新值时，进行检查（5,7）以查看该值是否已经存在于相应的存储器中（6,9）。 如果该检查的结果是肯定的，则在编码的隶属函数的情况下，编码的模糊推理指向这些函数的机制被改变，使得编码的模糊推理的指针被重定向到已经存在的隶属函数 存储。 在前提的操作数的情况下，最好根据对应的计算值对相应的备用存储（9）进行检查，当操作数的计算被发现为相应的 计算参数已经存在于相应的备用存储（9）中。 上述存储器（9）优选地被配置为在最上位置写入新的计算的操作数值的堆栈，使得被识别为已经存在的值可能返回到该最高位置的可能性。

公开(公告)号：EP0851342B1

公开(公告)日：2003-05-14

申请号：EP96830656.3

申请日：1996-12-27

Applicant: STMicroelectronics S.r.l.

Inventor： Pappalardo, Francesco ,  Arcidiacono, Liliana ,  Giacalone, Biagio ,  Di Bella, Dario

IPC: G06N7/02

CPC classification number: G06N7/04

公开(公告)号：EP1193645A1

公开(公告)日：2002-04-03

申请号：EP00830647.4

申请日：2000-10-02

Applicant: STMicroelectronics S.r.l.

Inventor： Pappalardo, Francesco ,  Giacalone, Biagio ,  Palano, Carmelo Marcello ,  Luzzi, Claudio ,  Grande, Francesca

IPC: G06N7/04 ,  G06F1/035

CPC classification number: G06F1/035 ,  G06N7/04

Abstract: The invention relates to a codifying and storing method for membership functions (MF) representing a membership degree (α) of fuzzy variables (x) defined within a universe of discourse (U.d.D.) which is discretized into a finite number of points (n). The membership functions (MF) are quantized into a finite number of levels (p) corresponding to a finite number of membership degrees, and are stored by means of a characteristic value of each sub-set of values of fuzzy variables (x) having for their image the same value of the membership degree (α) corresponding to one of said levels.
The invention further relates to a method for calculating the value of the membership degree of a fuzzy variable defined within a universe of discourse (U.d.D.) discretized into a finite number of points (n) with reference to a membership function (MF) thereof, as well as to a circuit (10) for calculating the membership degree (α) of a fuzzy variable (x) with reference to a membership function (MF) thereof.

Abstract translation: 本发明涉及一种用于隶属函数表示话语的宇宙（U.d.D.）所有其离散成有限数目的节点（n）的内定义模糊变量（X）的隶属度（阿尔法）一个编纂和存储方法（MF）。 隶属函数（MF）被量化成对应于一个有限数目的隶属度的有限数量的水平（P），并且借助于每个子集的模糊变量值具有用于（X）的特性值的存储 其图像的隶属程度（阿尔法）相同的值对应于所述级别中的一个。 本发明还涉及到用于计算话语（UDD）的宇宙中定义的模糊变量的隶属度的值的方法参照隶属函数（MF）离散成有限数目的节点（n）的物， 以及为参照隶属函数（MF）其计算模糊变量（x）的隶属度（阿尔法）的电路（10）。