WO2009097728A1 - 沥青混凝土搅拌设备配料误差的鲁棒控制方法 - Google Patents

沥青混凝土搅拌设备配料误差的鲁棒控制方法 Download PDF

Info

Publication number
WO2009097728A1
WO2009097728A1 PCT/CN2008/073306 CN2008073306W WO2009097728A1 WO 2009097728 A1 WO2009097728 A1 WO 2009097728A1 CN 2008073306 W CN2008073306 W CN 2008073306W WO 2009097728 A1 WO2009097728 A1 WO 2009097728A1
Authority
WO
WIPO (PCT)
Prior art keywords
output
weight
input
neuron
door
Prior art date
Application number
PCT/CN2008/073306
Other languages
English (en)
French (fr)
Inventor
Linfang Qian
Jian Jiang
Xiudong Shi
Changying Lai
Xianhui Wang
Longmiao Chen
Yadong Xu
Xin Deng
Yuanming Lu
Yan Lu
Guoyan Lu
Jianjun Ren
Zhenjun Meng
Original Assignee
Nanjing University Of Science And Technology
Wuxi Xitong Engineering Machinery Co., Ltd.
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing University Of Science And Technology, Wuxi Xitong Engineering Machinery Co., Ltd. filed Critical Nanjing University Of Science And Technology
Publication of WO2009097728A1 publication Critical patent/WO2009097728A1/zh

Links

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/0265Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
    • G05B13/027Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks only

Definitions

  • the invention belongs to a robust control method, in particular to a robust control method for batching error of asphalt concrete mixing equipment. Background technique
  • Asphalt mixture concrete mixing equipment is the key equipment for the construction of high-grade roads.
  • the quality of the mixture plays a decisive role in the quality of the road.
  • the error of the ingredients plays an important role.
  • the ingredients of the asphalt mixture mixing equipment account for more than 90% of the excess due to the change in the amount of flying material.
  • the current methods of ingredients are as follows:
  • Fixed value correction method also known as fixed impulse correction method. This correction method is related to various factors such as the amount of material in the warehouse, the stability of the door movement, and the composition change of the material. The batching error is greater than 10%. Meet the accuracy requirements.
  • the impulse self-correction method is the automatic correction or the impulse self-correction method.
  • the system automatically records the change of the fly material, and then superimposes an impulse automatic adjustment value on the fixed impulse value, and passes the superimposed impulse value. Controlling the early closing of the ingredient compartment door enables the batching system to follow changes in the fly stock. The method is only used when the amount of flying material fluctuates, but the fluctuation speed is not large, and the batching precision is high. Since the historical data is used or the correction value of this time is calculated based on the previous distribution values, it is often said that this time is the next time, which is behind the current material condition, so the real-time performance is poor.
  • the object of the present invention is to provide a robust control method for adaptively correcting the amount of advancement of the fly material and greatly improving the batching accuracy of the asphalt concrete mixing equipment.
  • the technical solution for achieving the object of the present invention is: a robust control method for the batching error of the asphalt concrete mixing device.
  • the weight of the weighing bin is the tare weight Wo at the initial stage, and after the door is opened, the material falls into the weighing scale by its own weight.
  • w 2 is the actual flyweight weight
  • the control process is as follows:
  • an input-output artificial neural network model adopt a three-layer structure, input a layer of neurons, represent the flow Q in front of the gate; five neurons in the middle layer; one neuron in the output layer, representing the actual amount of fly Y; A nonlinear model of the flow rate Q and the actual fly material amount Y in front of the closed door.
  • This is a black box model.
  • the dynamic prediction and learning phase that is, the prediction phase:
  • the value of the output neuron Y is represented by the forward calculation, and Y represents The amount of flying material, this kind of prediction is calculated periodically, and the amount of flying material Y is continuously predicted.
  • ⁇ + ⁇ - ⁇ 1 ⁇ 4 is equal to the set distribution value W of this time, the closing door signal is issued, and the forecasting stage is completed;
  • E J (F* - w 2 k )
  • Y k is the output calculated from the current weight based on the input of the kth learning sample, ie the flow Q k when the door is closed.
  • W 2 k is the actual amount of material spent on the kth learning sample.
  • the invention has the remarkable advantages: a nonlinear model of the flow rate and the amount of the fly material is established, and through continuous learning adjustment, the change of the external parameter can be adaptively compared with the linear model of the fixed parameter, on the asphalt
  • the mixing error of the mixing equipment of the mixing material has achieved better control effect, which can make the batching error reach ⁇ 2%, and improve the batching precision of the asphalt mixing equipment.
  • Figure 1 is a schematic diagram of batching control.
  • FIG. 2 is a schematic diagram of an artificial neural network model of the present invention. detailed description
  • the silo is filled with material, and the silo door controls the opening and closing of the silo.
  • the weight of the scale is the tare weight Wo.
  • the material falls into the weighing scale by its own weight.
  • the dynamic weight of the scale can be read and the door can be closed.
  • the static weight W can be read.
  • the weight of the material in the scale from the opening to the complete closing of the door W Wi+WrW where: W 2 is the weight of the flying material.
  • the setting value is set to W, it can be known from the above formula that the setting is required to close the door by W 2 before the set value. Because W 2 is related to the amount of material in the warehouse, the stability of the door movement, and the composition change of the material, W 2 changes drastically.
  • the value of W 2 is related to the flow rate value of the process of starting to close the material and completely falling into the scale.
  • the flow value cannot be considered to be equal to the flow value collected in real time before closing and is constant. They are different in timing and thus numerically. There are also differences, and the time when the material completely falls into the scale is also related to the flow, not a certain value, so if the linear model is used, there will be some error.
  • the amount of flying material is related to Q. This relationship is also time-varying. For example, the warehouse door gradually becomes slower due to mechanical reasons, resulting in a larger amount of flying material. Therefore, it is not appropriate to adopt a fixed model. How to solve this time-varying nonlinear model and find a robust control method is the key problem to be solved by the present invention.
  • the dynamic prediction and learning phase that is, the prediction phase:
  • the prediction phase According to the current real-time traffic Q as the value of the input neuron, according to the current weight of the artificial neural network model, the value of the output neuron Y is represented by the forward calculation, and Y represents The amount of flying material, this prediction is calculated by timing (set the scanning period, such as 20ms), and continuously predicts the amount of flying material Y. When ⁇ + ⁇ - ⁇ 1 ⁇ 4) is equal to the set matching value W of this time, it is closed. The warehouse door signal is completed in the forecast phase;
  • the learning uses the Back Propagation algorithm, which is the forward propagation of the input information and the inverse propagation of the error.
  • Back Propagation for a training sample, its feature vector, that is, the flow before closing the gate, is input into the neural network, and the forward propagation of the neural network is calculated to obtain an actual output Y k , and then the output is compared with the expected sample. The actual flying material amount W 2 k is compared. If there is a deviation, it is transferred to the back propagation process, and the deviation is returned from the original communication path.
  • the error is reduced; , then transfer to the forward propagation process, iteratively iterative, until the error is less than or equal to the allowable value, the learning ends. Therefore, learning training includes two processes of forward propagation of input vector and back propagation of error. It can be seen that the advantage of this three-layer feedforward network is that it has strong nonlinear mapping capability and flexible network structure.
  • the sensor output of the weighing scale is collected periodically, converted into real-time flow, and input into the input layer neuron of FIG. 1 as input quantity, and the output of the output layer neuron is obtained by forward calculation, and the input and output of the neuron are used in the calculation.
  • the output of the output layer neurons is the predicted amount of flying material. When the amount of flying material plus the weight of the material that has fallen into the scale is removed and the net weight is equal to the set ratio of this time, the closing door signal is issued. When all the flying materials fall into the weighing scale, the actual amount of flying material is taken as the output, and the 2 flow rate when the door is closed is used as an input to reform the learning sample.
  • the adjustment weight is such that w is the smallest, where Y k is the input quantity of the kth learning sample (the flow rate Q k when the door is closed ), and the output value calculated according to the current weight after inputting the artificial neural network, W k is the first The actual amount of material consumed for k learning samples. The predicted value of the fly in the next position is calculated based on the weight of this correction. It can be seen that the robust control method is divided into two dynamic stages of prediction and learning, and the non-linear relationship between the flow rate and the amount of fly material in the closed door is distributed in the connection weight of the neuron.

Landscapes

  • Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)
  • Preparation Of Clay, And Manufacture Of Mixtures Containing Clay Or Cement (AREA)

Description

沥青混凝土搅拌设备配料误差的鲁棒控制方法 技术领域
本发明属于鲁棒控制方法, 特别是一种沥青混凝土搅拌设备配料误差 的鲁棒控制方法。 背景技术
沥青混合料混凝土搅拌设备是高等级公路建设的关键设备, 混合料的 质量高低对路面质量起着决定性作用, 其中配料的误差起着重要的影响。 据文献报道, 沥青混合料搅拌设备配料因飞料量的变化导致超差占 90%以 上。 目前配料的方法有以下几种:
( 1 ) 定值修正法, 也称固定冲量修正法, 这种修正方法与仓内料的 多少、 仓门动作的稳定性、 物料的成分变化等多种因素有关, 配料误差大 于 10%, 无法满足精度要求。
(2) 冲量自修正法也即提前量自动修正或冲量自修正法, 系统自动 记录飞料的变化情况, 然后在固定冲量值上叠加一个冲量自动调整值, 通 过这个叠加后的冲量值控制配料仓门的提前关闭, 使配料系统能够跟随飞 料的变化。 该方法法仅在飞料量有波动, 但波动速度不大的情况下使用, 其配料精度较高。 由于采用的是历史数据或者说是根据上几次的配给值计 算本次的修正值, 即常说的本次补下次, 落后于当前物料状况, 所以实时 性差。
(3 ) 流量控制法。 在实际配料过程中经常会遇到飞料波动既大又快 的情况, 如在缺料时产生的飞料非常小, 由缺料突然转为正常时飞料又很 大, 此时系统使用正常情况下的冲量进行调整则会带来很大的配料误差。 如果 W2为飞料重量,则该方法中关门后的飞料量 W2 QS,其中 Q 为关 门时刻的流量, 单位是 kg / s; S为物料完全落入秤中的时间, 单位是 s。 该方法利用计算机足够高的采样速度, 不断采集变化的流量, 及时控制仓 门的关闭, 达到准确的配给。 由于 Q反映了当前时刻的物料流量, 排除了 料仓料位的影响, 料仓料多料少都能够及时修正本次的配料误差, 具有实 时性。 但这种方法认为关门后流量和物料完全落入秤中的时间为一定值, 采用的是线性模型, 由于式中的 Q是关门前实时采集的流量值, 与飞料完 全落入料仓的流量值是有时序上差异的, 因而在数值上也有差异, 而物料 完全落入秤中的时间也与流量有关系, 并非一定值, 因此该方法的线性模 型还是带来有一定的误差。
以上方法的介绍见 《筑路机械与施工矶械化》 期刊的 2000年第 4期 王志文的文章 "沥青混合料搅拌设备配料误差的最优控制"。 发明内容
本发明的目的在于提供一种对飞料量自适应修正关门的提前量, 大大 提高配料精度的沥青混凝土搅拌设备配料误差的鲁棒控制方法。
实现本发明目的的技术解决方案为: 一种沥青混凝土搅拌设备配料误 差的鲁棒控制方法, 料仓初始时秤重量为皮重 Wo, 打开仓门后, 物料靠自 重落入计量秤中, 此时可读出秤的动态重量 然后关闭仓门, 待秤完全 稳定后可读出静态重量 W, 那么仓门从开启到完全关闭后秤中的物料重量 W=W!+W2-W0, 式中: w2为实际飞料重量; 控制过程如下:
首先, 建立输入输出人工神经网络模型, 采用三层结构, 输入层一个 神经元, 代表关仓门前流量 Q ; 中间层五个神经元; 输出层一个神经元, 代表实际飞料量 Y ; 建立关仓门前流量 Q与实际飞料量 Y的非线性模型, 这 是一种黑箱模型,神经元的输入输出采用非线性函数 f(x)=l/l+e_x, X代表神 经元输入, f(x)代表神经元输出, 则 Vi=l/l+e_ui, Vi代表神经元的输出, Ui 代表神经元的输入, i=l〜5 ;
其次, 进行动态预测和学习阶段, 即预测阶段: 根据当前的实时流量 Q作为输入神经元的值, 根据人工神经网络模型当前的权值, 通过前向计 算得到输出神经元的值 Y, Y代表飞料量, 这种预测是定时计算的, 不断预 测飞料量 Y, 当\^+¥-\¼)等于本次的设定配给值 W时,发出关闭仓门信号, 预测阶段完成;
学习阶段: 预测阶段完成后, 得到采集的关仓门前流量 Q与实际飞料 量 W2这个最新采集样本对网络进行重新学习调整权值,使下式的配料误差 E最小:
E = J (F* - w2 k ) 12 式中 Yk是根据第 k个学习样本的输入量即关闭仓门时的流量 Qk,输入到 人工神经网络后根据当前权值计算出来的输出值, W2 k为第 k个学习样本的 实际飞料量。
本发明与现有技术相比, 其显著优点: 建立了流量和飞料量的非线性 模型, 并通过不断的学习调整, 与固定参数的线性模型相比能自适应外界 参数的变化, 对沥青混合料搅拌设备配料误差达到了更好的控制效果, 这 样可以使配料误差达到≤±2%, 提高了沥青混合料搅拌设备的配料精度。
下面结合附图对本发明作进一步详细描述。 附图说明
图 1是配料控制示意图。
图 2是本发明的人工神经网络模型示意图。 具体实施方式
如图 1所示, 料仓充满物料, 仓门可控制料仓的开闭。 初始时秤重量 为皮重 Wo, 打开仓门后, 物料靠自重落入计量秤中, 此时可读出秤的动态 重量 然后关闭仓门, 待秤完全稳定后可读出静态重量 W, 那么仓门从 开启到完全关闭后秤中的物料重量 W Wi+WrW 式中: W2为飞料重量。
若设定配给值为 W, 由上式可得知比设定值提前 W2关闭仓门即可达到 设定要求。 由于 W2与仓内料的多少、 仓门动作的稳定性、 物料的成分变化 等多种因索有关, 所以 W2是时刻剧烈变化的。
W2的值与开始关门到料完全落入秤中过程的料流量值有关,该流量值 不能认为等于关门前实时采集的流量值且为定值, 它们在时序上差异的, 因而在数值上也有差异, 而物料完全落入秤中的时间也与流量有关系, 并 非一定值, 因此如果采用线性模型会带来有一定的误差。 飞料量与 Q是有 关系的, 这种关系也是时变的, 比如仓门由于机械的原因渐渐反应变慢, 导致飞料量变大, 因此如果采用固定的模型也是不合适的。 如何解决这种 时变的非线性模型, 寻找一种鲁棒的控制方法是本发明重点要解决的问 题。
结合图 2, 本发明沥青混凝土搅拌设备配料误差的鲁棒控制方法的过 程如下, 首先, 建立输入输出人工神经网络模型, 采用三层结构, 输入层 一个神经元, 代表关仓门前流量 Q; 中间层五个神经元; 输出层一个神经 元,代表实际飞料量 Y;建立关仓门前流量 Q与实际飞料量 Y的非线性模型, 这是一种黑箱模型,神经元的输入输出采用非线性函数 f(x)=l/l+e_x, X代表 神经元输入, f(x)代表神经元输出, 则 Vi=l/l+e , ¼代表神经元的输出, Ui代表神经元的输入, i=l〜5 ;
其次, 进行动态预测和学习阶段, 即预测阶段: 根据当前的实时流量 Q作为输入神经元的值, 根据人工神经网络模型当前的权值, 通过前向计 算得到输出神经元的值 Y, Y代表飞料量, 这种预测是定时计算的(设定扫 描周期, 比如 20ms), 不断预测飞料量 Y, 当\^+¥-\¼)等于本次的设定配 给值 W时,发出关闭仓门信号,预测阶段完成;
学习阶段: 预测阶段完成后, 得到采集的关仓门前流量 Q与实际飞料 量 W2这个最新采集样本对网络进行重新学习调整权值,使下式的配料误差 E最小:
£ = J (F* - w2 k ) 12 式中 Yk是根据第 k个学习样本的输入量即关闭仓门时的流量 Qk,输入到 人工神经网络后根据当前权值计算出来的输出值, W2 k为第 k个学习样本的 实际飞料量。
其中学习采用 Back Propagation算法, 即输入信息的正向传播和误差的 反相传播。 在正向传播中, 对于一个训练样本, 将其特征向量, 即关仓门 前流量输入神经网络, 经过神经网络的前向传播计算, 得到一个实际输出 Yk, 然后将该输出与期望的样本输出实际飞料量 W2 k相比较, 若有偏差, 则转入反向传播过程, 将该偏差由原来的联络通路返回, 通过调整各层神 经元的联系权值, 使误差减小; 然后, 再转入正向传播过程, 反复迭代, 直到误差小于等于允许值, 学习才结束。 因此, 学习训练包括输入向量的 前向传播和误差的反向传播两个过程, 可见这个三层前馈网络的优点就是 具有很强的非线性映射能力和柔性的网络结构。
这种学习和预测是不断在线进行的, 一旦系统的参数出现变化, 人工 神经网络的权值会出现变化, 所以算法是鲁棒的。
实施例
定时采集计量秤的传感器输出, 转化成实时流量, 将其作为输入量定 时输入图 1的输入层神经元, 通过前向计算, 得到输出层神经元的输出, 在计算中神经元的输入输出采用非线性函数。 输出层神经元的输出就是预 测的飞料量, 当飞料量加上已落到秤中的料的重量再除去净重等于本次的 设定配给值时, 发出关闭仓门信号。 当飞料全部落入计量秤后, 将实际的 飞料量作为输出, 关闭仓门时的 2流量作为输入, 重新形成学习样
E = Y(Yk - w2 k ) 12
本调整权值, 使 w 最小, 其中 Yk是第 k个学习样本的输 入量 (关闭仓门时的流量 Qk)输入人工神经网络后根据当前权值计算出来的 输出值, Wk为第 k个学习样本的实际飞料量。 下一仓的飞料预测值根据本 次修正的权值来计算。 可见, 该鲁棒控制方法分为预测和学习两个动态的 阶段, 关仓门前流量与飞料量的非线性关系分布在神经元的连接权中。

Claims

权 力 要 求 书
1.一种沥青混凝土搅拌设备配料误差的鲁棒控制方法, 料仓初始时秤重量 为皮重 WQ, 打开仓门后, 物料靠自重落入计量秤中, 此时可读出秤的动态 重量 然后关闭仓门, 待秤完全稳定后可读出静态重量 w, 那么仓门从 开启到完全关闭后秤中的物料重量 W Wi+WrWo,式中: W2为实际飞料重 量; 控制过程如下:
首先, 建立输入输出人工神经网络模型, 采用三层结构, 输入层一个 神经元, 代表关仓门前流量 Q; 中间层五个神经元; 输出层一个神经元, 代表实际飞料量 Y; 建立关仓门前流量 Q与实际飞料量 Y的非线性模型, 这 是一种黑箱模型,神经元的输入输出采用非线性函数 f(x)=l/l+e—, X代表神 经元输入, f(x)代表神经元输出, 则 Vi=l/l+e— Ul, Vi代表神经元的输出, Ui 代表神经元的输入, i=l~5;
其次, 进行动态预测和学习阶段, 即预测阶段: 根据当前的实时流量 Q作为输入神经元的值, 根据人工神经网络模型当前的权值, 通过前向计 算得到输出神经元的值 Y, Y代表飞料量, 这种预测是定时计算的, 不断预 测飞料量 Y, 当W1+Y-WQ等于本次的设定配给值 W时,发出关闭仓门信号, 预测阶段完成;
学习阶段: 预测阶段完成后, 得到采集的关仓门前流量 Q与实际飞料 量 W2这个最新采集样本对网络进行重新学习调整权值,使下式的配料误差 E最小:
E =∑(Yk - w2 k ) 12 式中 Yk是根据第 k个学习样本的输入量即关闭仓门时的流量 Qk,输入到 人工神经网络后根据当前权值计算出来的输出值, W2 k为第 k个学习样本的 实际飞料量。
2.根据权利要求 1所述的沥青混凝土搅拌设备配料误差的鲁棒控制方法, 其特征在于学习阶段采用 Back Propagation算法, 即输入信息的正向传播和 误差的反相传播, 在正向传播中, 对于一个训练样本, 将其特征向量即关 仓门前流量输入神经网络, 经过神经网络的前向传播计算, 得到一个实际 输出 Yk, 然后将该输出与期望的样本输出实际飞料量 W2 k相比较, 若有偏 差, 则转入反向传播过程, 将该偏差由原来的联络通路返回, 通过调整各 层神经元的联系权值, 使误差减小; 然后, 再转入正向传播过程, 反复迭 代, 直到误差小于等于允许值, 学习才结束。
PCT/CN2008/073306 2008-02-04 2008-12-03 沥青混凝土搅拌设备配料误差的鲁棒控制方法 WO2009097728A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN 200810020530 CN101226377B (zh) 2008-02-04 2008-02-04 沥青混凝土搅拌设备配料误差的鲁棒控制方法
CN200810020530.7 2008-02-04

Publications (1)

Publication Number Publication Date
WO2009097728A1 true WO2009097728A1 (zh) 2009-08-13

Family

ID=39858417

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2008/073306 WO2009097728A1 (zh) 2008-02-04 2008-12-03 沥青混凝土搅拌设备配料误差的鲁棒控制方法

Country Status (2)

Country Link
CN (1) CN101226377B (zh)
WO (1) WO2009097728A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102053628A (zh) * 2009-10-27 2011-05-11 北京航空航天大学 基于神经网络的伺服控制系统及方法
CN111539111A (zh) * 2020-04-27 2020-08-14 上海建工集团股份有限公司 一种超高层建筑混凝土泵送性能测控系统

Families Citing this family (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101226377B (zh) * 2008-02-04 2010-11-24 南京理工大学 沥青混凝土搅拌设备配料误差的鲁棒控制方法
CN102636245B (zh) * 2012-04-23 2013-11-06 中联重科股份有限公司 一种物料的称重测量方法、装置及称重测量系统
CN104932257B (zh) * 2015-05-15 2017-07-28 重庆大学 一种砖厂动态配料系统的鲁棒自适应控制方法
CN105974791B (zh) * 2016-06-20 2019-01-18 广州大学 一种混凝土配合比优化系统控制方法
CN107661728B (zh) * 2017-09-19 2019-09-24 中国计量大学 基于变速率学习的直落式物料配料装置及其控制器
CN107601064B (zh) * 2017-09-19 2019-09-24 中国计量大学 基于神经网络的直落失重式物料下料机及其控制器
CN107715727B (zh) * 2017-09-19 2019-10-25 中国计量大学 螺杆式多组份物料配料装置及其控制器
CN107694469B (zh) * 2017-09-19 2019-09-24 中国计量大学 基于变速率学习的直落式多组份物料配料方法
CN108002062B (zh) * 2017-09-19 2019-10-18 中国计量大学 基于神经网络的螺杆失重式物料下料方法
CN110697100B (zh) * 2017-09-19 2021-04-02 中国计量大学 直落式多组份物料下料装置控制器
CN110697449B (zh) * 2017-09-19 2021-07-02 中国计量大学 一种基于神经网络的螺杆失重式物料下料机控制器
CN107697660B (zh) * 2017-09-19 2019-11-29 中国计量大学 基于机器学习的螺杆式物料配料机控制方法
CN107684847B (zh) * 2017-09-19 2019-10-25 中国计量大学 螺杆式多组份物料配料方法
CN107673083B (zh) * 2017-09-19 2019-07-26 中国计量大学 基于变速率学习的螺杆式物料下料装置及其控制器
CN107741695B (zh) * 2017-09-19 2020-05-05 中国计量大学 基于机器学习的直落式物料下料机控制方法
CN110697448B (zh) * 2017-09-19 2021-07-02 中国计量大学 一种基于机器学习的螺杆式物料配料机控制器
CN107544252B (zh) * 2017-09-19 2020-08-11 中国计量大学 基于机器学习的直落式物料下料机控制器
CN107684846B (zh) * 2017-09-19 2019-10-11 中国计量大学 直落式多组份物料下料方法
CN108246201B (zh) * 2017-12-27 2021-03-16 云南云天化农业科技股份有限公司 一种用于配肥设备的控制方法、控制装置及配肥设备
CN109554992B (zh) * 2018-11-12 2021-02-26 福建省铁拓机械股份有限公司 一种冷料仓自动调节方法
CN112551084B (zh) * 2020-10-22 2022-07-29 沈阳建筑大学 一种预制构件混凝土布料的螺旋输送量的分析方法
CN116757253B (zh) * 2023-07-03 2024-06-11 浙江永贵博得交通设备有限公司 一种轨道交通智能自动学习开关门算法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH07227832A (ja) * 1994-02-21 1995-08-29 Tokyo Electric Power Co Inc:The コンクリートの品質管理方法及び装置
CN1299082A (zh) * 1999-12-07 2001-06-13 广州大学 Pid神经网络控制器
US20020019674A1 (en) * 2000-06-08 2002-02-14 Jui-Ming Liang Intelligent control method for injection machine
CN1851571A (zh) * 2006-03-09 2006-10-25 上海工程技术大学 物料平衡智能控制系统
CN101226377A (zh) * 2008-02-04 2008-07-23 南京理工大学 沥青混凝土搅拌设备配料误差的鲁棒控制方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH07227832A (ja) * 1994-02-21 1995-08-29 Tokyo Electric Power Co Inc:The コンクリートの品質管理方法及び装置
CN1299082A (zh) * 1999-12-07 2001-06-13 广州大学 Pid神经网络控制器
US20020019674A1 (en) * 2000-06-08 2002-02-14 Jui-Ming Liang Intelligent control method for injection machine
CN1851571A (zh) * 2006-03-09 2006-10-25 上海工程技术大学 物料平衡智能控制系统
CN101226377A (zh) * 2008-02-04 2008-07-23 南京理工大学 沥青混凝土搅拌设备配料误差的鲁棒控制方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
WANG, Z.: "Optimum Control of Aggregate Feeding Error for Asphalt Mixing Plant", ROAD MACHINERY & CONSTRUCTION MECHANIZATION, vol. 174, 31 December 2000 (2000-12-31), pages 7 - 8 *
YANG, S., ET AL.: "Application of Neural Network in Cement Raw Materials Blending System", INDUSTRY AND MINE AUTOMATION, vol. 5, 31 October 2004 (2004-10-31), pages 7 - 9 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102053628A (zh) * 2009-10-27 2011-05-11 北京航空航天大学 基于神经网络的伺服控制系统及方法
CN102053628B (zh) * 2009-10-27 2014-10-22 北京航空航天大学 基于神经网络的伺服控制系统及方法
CN111539111A (zh) * 2020-04-27 2020-08-14 上海建工集团股份有限公司 一种超高层建筑混凝土泵送性能测控系统
CN111539111B (zh) * 2020-04-27 2024-04-05 上海建工集团股份有限公司 一种超高层建筑混凝土泵送性能测控系统

Also Published As

Publication number Publication date
CN101226377A (zh) 2008-07-23
CN101226377B (zh) 2010-11-24

Similar Documents

Publication Publication Date Title
WO2009097728A1 (zh) 沥青混凝土搅拌设备配料误差的鲁棒控制方法
CN102615710B (zh) 一种配料计量的方法、配重控制器、系统及混凝土搅拌站
CN102096373B (zh) 基于增量改进bp神经网络的微波干燥pid控制方法
CN104846306B (zh) 一种镀锌厚度控制系统及方法
CN101995822A (zh) 一种大时滞系统的灰色自抗扰控制方法
CN101957598A (zh) 一种大时滞系统的灰色无模型控制方法
CN106354017A (zh) 一种稀土萃取分离过程组分含量区间控制方法
CN106682735A (zh) 基于pid调节的bp神经网络算法
Jiang et al. Real-time moisture control in sintering process using offline–online NARX neural networks
CN106484977A (zh) 高拱坝施工进度自适应仿真分析方法
CN102354119A (zh) 一种设施农业环境智能控制方法
CN109960149A (zh) 用于pH中和过程的自抗扰控制参数整定方法
Aggarwal et al. Prediction of compressive strength of self-compacting concrete with fuzzy logic
CN201476905U (zh) 神经网络pid控温的热电偶自动检定系统
CN103679268A (zh) 高炉炉渣粘度预报方法
Yang et al. Modeling and optimal-setting control of blending process in a metallurgical industry
CN104597753A (zh) 一种同步碎石封层车沥青碎石撒布智能控制方法及装置
CN110095985B (zh) 一种观测器设计方法和抗干扰控制系统
CN109829177A (zh) 一种多工况环境下球磨机溢流粒度分布软测量方法
CN107741695A (zh) 基于机器学习的直落式物料下料机控制方法
CN109870986A (zh) 一种基于神经网络和数据驱动的搅拌反应釜在线控制方法
Duffie et al. Analysis of Lead-Time Regulation in an Autonomous Work System
CN106096761A (zh) 一种基于神经网络的建筑负荷预测方法及装置
CN101414157A (zh) 基于遗传算法和迭代学习法的智能称重控制方法
CN115313411B (zh) 一种电网无功优化控制方法和装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 08872172

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 08872172

Country of ref document: EP

Kind code of ref document: A1