[go: up one dir, main page]

CN109211233B - Elevator motion detection and abnormal position parking judgment method based on acceleration sensor - Google Patents

Elevator motion detection and abnormal position parking judgment method based on acceleration sensor Download PDF

Info

Publication number
CN109211233B
CN109211233B CN201811112016.6A CN201811112016A CN109211233B CN 109211233 B CN109211233 B CN 109211233B CN 201811112016 A CN201811112016 A CN 201811112016A CN 109211233 B CN109211233 B CN 109211233B
Authority
CN
China
Prior art keywords
elevator
elevator car
floor
car
estimated
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.)
Active
Application number
CN201811112016.6A
Other languages
Chinese (zh)
Other versions
CN109211233A (en
Inventor
朱培逸
孙振
吕岗
徐本连
施健
鲁明丽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changshu Institute of Technology
Original Assignee
Changshu Institute of Technology
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 Changshu Institute of Technology filed Critical Changshu Institute of Technology
Priority to CN201811112016.6A priority Critical patent/CN109211233B/en
Publication of CN109211233A publication Critical patent/CN109211233A/en
Application granted granted Critical
Publication of CN109211233B publication Critical patent/CN109211233B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
    • G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
    • G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/20—Instruments for performing navigational calculations

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Navigation (AREA)
  • Indicating And Signalling Devices For Elevators (AREA)

Abstract

本发明公开了一种基于加速度传感器的电梯运动检测及异常位置停车判断，按照如下步骤进行：步骤1.电梯轿厢行驶距离估计，位于电梯轿厢上的三轴加速度传感器采集电梯的运动参数，通过扩展卡尔曼滤波算法，估计每一趟旅程中即两次停车之间电梯的行驶距离，返回电梯在一趟行驶过程中的位移估计值及方差；步骤2.电梯轿厢停层及故障判断，基于步骤1的位移估计值及方差结果，结合SLAM算法，在设置基准点的基础上估计出电梯轿厢的位置和楼层的高度；步骤3.通过位置估计判断电梯是否发生异常停车故障。本发明利用三轴加速度计实现对电梯运动状态的检测，利用扩展卡尔曼滤波方法对电梯状态进行估计，可以快速得到电梯轿厢的位移信息。

Figure 201811112016

The invention discloses an elevator motion detection and abnormal position parking judgment based on an acceleration sensor, which is carried out according to the following steps: Step 1. Estimating the travel distance of an elevator car, a three-axis acceleration sensor located on the elevator car collects the motion parameters of the elevator, Through the extended Kalman filter algorithm, the travel distance of the elevator between two stops in each trip is estimated, and the estimated displacement and variance of the elevator during one trip are returned; Step 2. Elevator car floor stop and fault judgment , based on the displacement estimation value and variance result of step 1, combined with the SLAM algorithm, on the basis of setting the reference point, the position of the elevator car and the height of the floor are estimated; step 3. Determine whether the elevator has abnormal parking failure through position estimation. The invention utilizes the three-axis accelerometer to realize the detection of the elevator motion state, utilizes the extended Kalman filtering method to estimate the elevator state, and can quickly obtain the displacement information of the elevator car.

Figure 201811112016

Description

基于加速度传感器的电梯运动检测及异常位置停车判断方法Elevator motion detection and abnormal position parking judgment method based on acceleration sensor

技术领域technical field

本发明属于电梯运动检测领域，更具体的涉及一种基于加速度传感器的电梯运动检测及异常位置停车判断方法。The invention belongs to the field of elevator motion detection, and more particularly relates to a method for elevator motion detection and abnormal position parking judgment based on an acceleration sensor.

背景技术Background technique

电梯轿厢位置精确估计对于电梯正常运行具有极为重要的意义。当电梯异常停层时，电梯与楼层未能对准的情况下开门，会发生人员坠梯的危险。目前电梯轿厢准确停层检测方法，都是基于电梯与楼层信号相互呼应的原理。非常依赖信号的稳定和准确，并且需要安装大量检测信号发出和接收装置(每个楼层都需要安装)，维护成本高，适用型差。当电梯由于急停、系统断电或者曳引绳与曳引轮之间发生滑移等情况导致电梯位置丢失，如果无法确认电梯轿厢位置，则电梯将无法正常工作。Accurate estimation of the elevator car position is of great significance for the normal operation of the elevator. When the elevator stops abnormally, the door will open if the elevator and the floor are not aligned, and there will be a danger of people falling. At present, the accurate floor stop detection methods of the elevator car are all based on the principle that the elevator and floor signals echo each other. It is very dependent on the stability and accuracy of the signal, and needs to install a large number of detection signal sending and receiving devices (every floor needs to be installed), the maintenance cost is high, and the applicability is poor. When the elevator loses its position due to emergency stop, system power failure, or slippage between the traction rope and the traction sheave, if the position of the elevator car cannot be confirmed, the elevator will not work normally.

电梯异常位置停车检测的一个重要的方面就是确定电梯在哪个高度，并检测是否不平层、蹲底或冲顶。通常可以通过卫星定位系统(GPS)为用户提供高度(海拔)信息，但是这种高度判断在民用产品中误差较大，只适用于跨度较大的高度判断，不能用于室内跨度小的高度判断。另外，由于建筑外墙对卫星信号的屏蔽作用，在室内卫星定位系统的使用极为困难。An important aspect of elevator parking detection in abnormal locations is to determine what height the elevator is at, and to detect whether it is uneven, squatting, or topping. Usually, the altitude (altitude) information can be provided to the user through the satellite positioning system (GPS), but this kind of altitude judgment has a large error in civilian products, and is only suitable for the altitude judgment with a large span, and cannot be used for indoor altitude judgment with a small span . In addition, due to the shielding effect of building exterior walls on satellite signals, it is extremely difficult to use indoor satellite positioning systems.

本发明试图提供室内尤其是电梯井内的轿厢定位系统。该系统通过电梯运动检测系统，采集轿厢的运动状态等参数，来判断其所处的位置，并根据运动结果来确定电梯是否发生异常位置停车故障。运动检测系统中加速度计可以全方位感知物体的运动状态。通过检测加速度计各方向的受力，可以判断轿厢的运动方向，以及加速度的变化幅值和移动的距离(从静止开始)，最终获得电梯轿厢位置精确估计。目前还没有基于加速度计的电梯异常停层故障的解决方案。The present invention seeks to provide a car positioning system indoors, especially in elevator shafts. The system uses the elevator motion detection system to collect parameters such as the motion state of the car to determine its position, and to determine whether the elevator has a parking failure at an abnormal position according to the motion results. The accelerometer in the motion detection system can sense the motion state of the object in all directions. By detecting the force in each direction of the accelerometer, the moving direction of the car, as well as the change amplitude of the acceleration and the moving distance (starting from a standstill) can be judged, and finally an accurate estimation of the elevator car position can be obtained. At present, there is no accelerometer-based solution for the abnormal landing failure of elevators.

发明内容SUMMARY OF THE INVENTION

1、发明目的。1. The purpose of the invention.

本发明利用加速度计获取电梯的运动状态，将电梯在垂直方向上的加速度作为输入，通过扩展卡尔曼滤波估计每一趟旅程中(两次停车之间)电梯的行驶距离，再通过即时定位与地图构建(Simultaneous Localization and Mapping,SLAM)算法构建电梯行驶的轨迹，结合数据关联实现对电梯所处楼层的估计，并对电梯是否异常停层给出判断。The invention uses the accelerometer to obtain the motion state of the elevator, takes the acceleration of the elevator in the vertical direction as the input, estimates the travel distance of the elevator in each journey (between two stops) through extended Kalman filtering, and then uses the real-time positioning and The Simultaneous Localization and Mapping (SLAM) algorithm constructs the trajectory of the elevator, combines the data association to estimate the floor of the elevator, and judges whether the elevator stops abnormally.

2、本发明所采用的技术方案。2. The technical solution adopted in the present invention.

本发明公开了一种基于加速度传感器的电梯运动检测及异常位置停车判断方法，按照如下步骤进行:The invention discloses a method for elevator motion detection and abnormal position parking judgment based on an acceleration sensor, which is carried out according to the following steps:

步骤1.电梯轿厢行驶距离估计Step 1. Elevator car travel distance estimation

位于电梯轿厢上的三轴加速度传感器采集电梯的运动参数，通过扩展卡尔曼滤波算法，估计每一趟旅程中即两次停车之间电梯的行驶距离，返回电梯在一趟行驶过程中的位移估计值及方差；The three-axis acceleration sensor located on the elevator car collects the motion parameters of the elevator, and through the extended Kalman filter algorithm, the travel distance of the elevator between two stops in each trip is estimated, and the displacement of the elevator during one trip is returned. Estimates and variances;

步骤2.电梯轿厢停层及故障判断Step 2. Elevator car floor stop and fault judgment

基于步骤1的位移估计值及方差结果，结合SLAM算法，在设置基准点的基础上估计出电梯轿厢的位置和楼层的高度；Based on the displacement estimation value and variance result of step 1, combined with the SLAM algorithm, the position of the elevator car and the height of the floor are estimated on the basis of setting the reference point;

步骤3.通过位置估计判断电梯是否发生异常停车故障；所述的电梯轿厢行驶距离估计按照如下步骤进行：Step 3. Determine whether the elevator has an abnormal parking fault through position estimation; the estimation of the travel distance of the elevator car is carried out according to the following steps:

11)在电梯行驶距离估计中，电梯轿厢在连续两次停车(l-1,l)间移动距离为δp(l)和运动速度，通过横向和纵向速度为零的约束条件下辅助惯性导航即可估计得到，首先假设在t时刻，电梯的状态向量为xd(t)，其定义为11) In the estimation of the travel distance of the elevator, the moving distance of the elevator car between two consecutive stops (l-1,l) is δp(l) and the moving speed, and the inertial navigation is assisted under the constraint that the lateral and longitudinal speeds are zero. It can be estimated, first assume that at time t, the state vector of the elevator is x d (t), which is defined as

Figure GDA0002600768420000021
Figure GDA0002600768420000021

其中，d(t),s(t),δuz(t)分别表示电梯最后一次停车后运动的距离、轿厢的速度和垂直加速度测量偏差，包括地球引力造成的偏移，s(t-1)为辅助状态变量；Among them, d(t), s(t), δu z (t) represent the distance traveled by the elevator after the last stop, the speed of the car and the measurement deviation of vertical acceleration, including the offset caused by the earth's gravity, s(t- 1) is an auxiliary state variable;

12)在一次运动中，根据零速辅助惯性导航系统，可以定义时刻t电梯轿厢的运动状态矢量为：12) In a motion, according to the zero-speed auxiliary inertial navigation system, the motion state vector of the elevator car at time t can be defined as:

Figure GDA0002600768420000022
Figure GDA0002600768420000022

其中in

Figure GDA0002600768420000023
Figure GDA0002600768420000023

xd(t-1)是从上次停止开始行驶的距离，

Figure GDA0002600768420000031
是电梯轿厢测量的加速度，Δt为采样间隔，w(t)过程噪声，高斯白噪声，其协方差
Figure GDA0002600768420000032
其中
Figure GDA0002600768420000033
表示加速度计噪声方差，
Figure GDA0002600768420000034
表示加速度计偏差模型的噪声方差；x d (t-1) is the distance traveled since the last stop,
Figure GDA0002600768420000031
is the acceleration measured by the elevator car, Δt is the sampling interval, w(t) process noise, Gaussian white noise, and its covariance
Figure GDA0002600768420000032
in
Figure GDA0002600768420000033
represents the accelerometer noise variance,
Figure GDA0002600768420000034
represents the noise variance of the accelerometer bias model;

13)由于电梯轿厢的位置和运动状态只用加速度计来测量估计，因此无法为状态空间模型制定传统的量测更新方程；但是，电梯轿厢一般都是静止或者匀速运动，这两种状态都可以通过加速度计检测出来；假设已知电梯轿厢静止或者匀速运动的时间，那么状态空间的伪量测模型如下：13) Since the position and motion state of the elevator car are only measured and estimated by the accelerometer, it is impossible to formulate a traditional measurement update equation for the state space model; however, the elevator car is generally stationary or moving at a uniform speed, and these two states can be detected by the accelerometer; assuming that the time when the elevator car is stationary or moving at a constant speed is known, the pseudo-measurement model of the state space is as follows:

Figure GDA0002600768420000035
Figure GDA0002600768420000035

其中

Figure GDA0002600768420000036
Figure GDA0002600768420000037
和
Figure GDA0002600768420000038
分别是匀速运动和静止时的观测噪声，即伪量测模型为电梯运动估计提供了两个不同的更新模型：零速更新模型和匀速更新模型；in
Figure GDA0002600768420000036
Figure GDA0002600768420000037
and
Figure GDA0002600768420000038
They are the observation noise when moving at a constant speed and at rest, that is, the pseudo-measurement model provides two different update models for elevator motion estimation: a zero-speed update model and a constant-speed update model;

14)电梯轿厢只在有限范围内垂直方向移动，简化模型将电梯轿厢的运动均视为匀速直线运动，静止是一种特殊的匀速直线运动，同时三轴加速度传感器测量的加速度也取行程时间内的均值，并将电梯轿厢开始加速或减速的问题转化为电梯轿厢加速度均值局部变化的问题；14) The elevator car only moves in the vertical direction within a limited range. The simplified model regards the movement of the elevator car as a uniform linear motion. Rest is a special uniform linear motion. At the same time, the acceleration measured by the three-axis acceleration sensor is also taken as the stroke. The average value in time, and the problem of the elevator car starting to accelerate or decelerate is transformed into the problem of the local change of the average value of the elevator car acceleration;

15)根据状态空间矢量建立基于扩展卡尔曼滤波器的轿厢行驶距离估计和轿厢运动状态估计，将三轴加速度计测得平均加速度

Figure GDA0002600768420000039
作为算法的输入；15) Establish car travel distance estimation and car motion state estimation based on extended Kalman filter according to the state space vector, and measure the average acceleration of the three-axis accelerometer
Figure GDA0002600768420000039
as input to the algorithm;

16)电梯处于均匀线性运动时，通过计算将零速度假设拟合到观测数据后获得归一化预测误差ξd的大小，判断电梯轿厢是静止还是匀速运动；设置阈值γd，确定电梯轿厢的状态，对观测模型进行更新：如果ξd＞γd，选择匀速测量更新；如果ξd＜γd，选择零速度测量更新；16) When the elevator is in uniform linear motion, the normalized prediction error ξ d is obtained by fitting the zero-speed assumption to the observation data, and the elevator car is determined to be stationary or moving at a constant speed; set the threshold γ d to determine the elevator car According to the state of the car, update the observation model: if ξ d > γ d , select uniform speed measurement update; if ξ d < γ d , select zero speed measurement update;

17)当算法检测到电梯轿厢停止运动并且行驶的距离超过了阈值γδp，就会输出电梯轿厢行驶的距离估计值

Figure GDA00026007684200000310
和其方差估计值
Figure GDA00026007684200000311
并且对系统状态矢量进行更新，并返回更新值，在下一次运动时重复15)-17)过程。17) When the algorithm detects that the elevator car stops moving and the distance traveled exceeds the threshold γδp , it outputs an estimate of the distance traveled by the elevator car
Figure GDA00026007684200000310
and its variance estimate
Figure GDA00026007684200000311
And update the system state vector, and return the updated value, repeat the process 15)-17) in the next movement.

更进一步，步骤15)中Further, step 15) in

首先，通过扩展卡尔曼滤波算法对状态矢量估计值

Figure GDA0002600768420000041
和它的协方差矩阵
Figure GDA0002600768420000042
进行更新迭代，具体迭代步骤分为预测与更新：First, the state vector is estimated by the extended Kalman filter algorithm.
Figure GDA0002600768420000041
and its covariance matrix
Figure GDA0002600768420000042
The update iteration is performed, and the specific iteration steps are divided into prediction and update:

①预测①Prediction

Figure GDA0002600768420000043
Figure GDA0002600768420000043

Figure GDA0002600768420000044
Figure GDA0002600768420000044

②更新②Update

Figure GDA0002600768420000045
Figure GDA0002600768420000045

Figure GDA0002600768420000046
Figure GDA0002600768420000046

Figure GDA0002600768420000047
Figure GDA0002600768420000047

然后，通过均匀线性运动检测器判断电梯轿厢是否是均匀线性运动，如果是就继续执行迭代更新算法，如果不是就进行下一次行程的计算。Then, the uniform linear motion detector is used to judge whether the elevator car is in uniform linear motion, if so, continue to execute the iterative update algorithm, and if not, perform the calculation of the next trip.

更进一步，所述的步骤2电梯轿厢停层及故障判断按照如下步骤进行：Further, in the step 2, the elevator car stops and the fault judgment is carried out according to the following steps:

21)通过步骤1获取轿厢行驶距离估计值

Figure GDA0002600768420000048
和方差估计值
Figure GDA0002600768420000049
在此基础上，通过SLAM算法可以实现电梯轿厢位置和停靠楼层估计，同时通过估计的轿厢位置，检测电梯是否存在故障；21) Obtain the estimated value of the car travel distance through step 1
Figure GDA0002600768420000048
and variance estimates
Figure GDA0002600768420000049
On this basis, the SLAM algorithm can be used to estimate the elevator car position and the parking floor, and at the same time, through the estimated car position, it is possible to detect whether there is a fault in the elevator;

22)引入新的状态矢量

Figure GDA00026007684200000410
其中，p(l)为电梯轿厢在第l趟行驶后的位置，δp(l)表示电梯轿厢在连续两次停车(l-1,l)间的移动距离；m(i)为第i层楼的高度，M为楼层总数；由于每层楼的高度是恒定的，状态矢量描述为：22) Introduce a new state vector
Figure GDA00026007684200000410
Among them, p(l) is the position of the elevator car after the lth trip, δp(l) represents the moving distance of the elevator car between two consecutive stops (l-1,l); m (i) is the first The height of the i floor, M is the total number of floors; since the height of each floor is constant, the state vector is described as:

Figure GDA00026007684200000411
Figure GDA00026007684200000411

其中e1为单位阵，

Figure GDA00026007684200000412
是电梯轿厢行驶距离估计值的误差，假定误差ωs(l)为零均值、不相关，且其方差为
Figure GDA00026007684200000413
where e1 is the identity matrix,
Figure GDA00026007684200000412
is the error of the estimated value of the travel distance of the elevator car, assuming that the error ω s (l) is zero mean, uncorrelated, and its variance is
Figure GDA00026007684200000413

23)假设电梯轿厢第l次行驶所停的楼层i已知的，则对应的伪观测模型为：23) Assuming that the floor i where the elevator car stops at the lth time is known, the corresponding pseudo-observation model is:

Figure GDA00026007684200000414
Figure GDA00026007684200000414

其中

Figure GDA0002600768420000051
vs(l)是伪测量误差，由于电梯控制系统不完善该伪测量误差是电梯轿厢停靠点位置和楼层高度之间微小偏差，该伪测量误差假定为零均值、不相关，伪测量误差vs(l)的方差为
Figure GDA0002600768420000052
假如楼层和电梯轿厢停靠点是已知，则通过基于扩展卡尔曼滤波的SLAM跟踪电梯轿厢位置、估计楼层高度；in
Figure GDA0002600768420000051
v s (l) is the pseudo-measurement error. Due to the imperfect elevator control system, the pseudo-measurement error is a small deviation between the elevator car stop position and the floor height. The pseudo-measurement error is assumed to be zero mean and irrelevant. The pseudo-measurement error The variance of v s (l) is
Figure GDA0002600768420000052
If the floor and elevator car stops are known, the elevator car position is tracked and the floor height is estimated by SLAM based on extended Kalman filtering;

24)在实际工作环境中，楼层总数M和电梯轿厢的行驶顺序是未知的；假如电梯轿厢初始状态估计

Figure GDA0002600768420000053
和相应的协方差矩阵
Figure GDA0002600768420000054
是有效的，则通过基于最大似然数的数据关联实现电梯轿厢位置与估计楼层高度间的关联，该数据关联选择使归一化预测误差最小的观测模型，故通过最小化函数即可获得电梯轿厢所处的楼层高度：24) In the actual working environment, the total number of floors M and the travel sequence of the elevator car are unknown; if the initial state of the elevator car is estimated
Figure GDA0002600768420000053
and the corresponding covariance matrix
Figure GDA0002600768420000054
is valid, then the association between the elevator car position and the estimated floor height is realized through the data association based on the maximum likelihood. The data association selects the observation model that minimizes the normalized prediction error, so it can be obtained by minimizing the function. Floor height of the elevator car:

Figure GDA0002600768420000055
Figure GDA0002600768420000055

25)要完成上述24)中所说的步骤，必须要有理想的初始值；获取理想初始状态的方法，通过让电梯轿厢从底层依次运动到最高层，每层至少停留一次；25) To complete the steps mentioned in the above 24), there must be an ideal initial value; the method for obtaining the ideal initial state is to allow the elevator car to move from the bottom floor to the highest floor in sequence, and stay at least once on each floor;

26)基于扩展卡尔曼滤波器的SLAM算法，将第1阶段获得的电梯轿厢行驶的距离估计值

Figure GDA0002600768420000056
和距离估计误差的方差估计值
Figure GDA0002600768420000057
作为输入，对状态向量
Figure GDA0002600768420000058
和状态协方差
Figure GDA0002600768420000059
进行更新迭代；26) Based on the SLAM algorithm of the extended Kalman filter, the estimated value of the distance traveled by the elevator car obtained in the first stage
Figure GDA0002600768420000056
and the variance estimate of the distance estimation error
Figure GDA0002600768420000057
As input, for the state vector
Figure GDA0002600768420000058
and state covariance
Figure GDA0002600768420000059
update iteratively;

27)计算归一化预测误差

Figure GDA00026007684200000510
并通过最小化函数求出电梯轿厢所处的楼层i；27) Calculate the normalized prediction error
Figure GDA00026007684200000510
And find the floor i where the elevator car is located by minimizing the function;

28)比较归一化误差ξs与假定阈值γs的大小，如果ξs＞γs，则表明估计误差过大，故电梯轿厢运行异常，此时状态向量不进行量测更新，并且报警提示电梯轿厢异常停靠故障；如果误差ξs＜γs，则证明电梯轿厢运行正常，对状态向量进行量测更新，并返回更新的值，进入下一轮行驶过程并重复步骤26)-28)。28) Compare the size of the normalized error ξ s and the assumed threshold γ s , if ξ s >γ s , it means that the estimation error is too large, so the elevator car runs abnormally, at this time, the state vector is not measured and updated, and an alarm is issued Prompt the elevator car to stop abnormally; if the error ξ s < γ s , it proves that the elevator car is running normally, the state vector is measured and updated, and the updated value is returned, and the next round of driving process is entered and step 26)- 28).

更进一步，所述的步骤26)，具体迭代步骤分为预测与更新：Further, described step 26), specific iterative steps are divided into prediction and update:

①预测①Prediction

Figure GDA0002600768420000061
Figure GDA0002600768420000061

Figure GDA0002600768420000062
Figure GDA0002600768420000062

②更新②Update

Figure GDA0002600768420000063
Figure GDA0002600768420000063

Figure GDA0002600768420000064
Figure GDA0002600768420000064

Figure GDA0002600768420000065
Figure GDA0002600768420000065

更进一步，检测电梯是否存在故障，故障包括不平层、蹲底或冲顶。Going a step further, detect whether there is a fault in the elevator, including uneven floors, squatting or topping.

3、本发明所产生的技术效果。3. The technical effect produced by the present invention.

(1)本发明利用三轴加速度计实现对电梯运动状态的检测，利用扩展卡尔曼滤波方法对电梯状态进行估计，可以快速得到电梯轿厢的位移信息。(1) The present invention uses a three-axis accelerometer to detect the motion state of the elevator, and uses the extended Kalman filtering method to estimate the elevator state, so that the displacement information of the elevator car can be quickly obtained.

(2)本发明利用三轴加速度计实现电梯楼层定位，无需电梯与楼层进行信息交互，成本低，适用型高。(2) The present invention utilizes the three-axis accelerometer to realize elevator floor positioning, does not require information exchange between the elevator and the floor, has low cost, and is highly applicable.

(3)本发明利用扩展卡尔曼滤波和SLAM方法，相比其他的方法，只需要让电梯遍历一次获取楼层、楼层层高等数据，即可进行后期电梯运动状态估计和停层异常检测，工作效率大大提高。(3) The present invention utilizes the extended Kalman filter and SLAM method. Compared with other methods, the elevator only needs to traverse the elevator once to obtain the data of floors and floors, and then the elevator motion state estimation and floor stop abnormality detection can be performed in the later stage. Greatly improve.

附图说明Description of drawings

图1为本发明电梯轿厢行驶距离估计流程图。Fig. 1 is a flow chart of estimating the travel distance of the elevator car according to the present invention.

图2为本发明电梯轿厢停层及故障判断流程图。Fig. 2 is a flowchart of the elevator car floor stop and fault judgment according to the present invention.

图3为具体实施方案中对电梯楼层定位估计图。Fig. 3 is an estimation diagram of elevator floor positioning in a specific embodiment.

具体实施方式Detailed ways

实施例Example

本发明提出一种基于三轴加速度传感器的电梯运动检测方法，包括三轴加速度传感器、数据线、上位机，将三轴加速度传感器放置于电梯轿厢上，在电梯运行时采集电梯的运动参数，如电梯运动方向、速度、加速度，将加速度计采集到的参数作为输入，通过扩展卡尔曼滤波算法，估计每一趟旅程中(两次停车之间)电梯的行驶距离。并返回电梯在一趟行驶过程中的位移估计值及方差。再结合SLAM算法，估计出电梯轿厢的位置和楼层的高度(在设置基准点的基础上)。最后通过位置估计判断电梯是否发生异常停车故障。The invention proposes an elevator motion detection method based on a three-axis acceleration sensor, comprising a three-axis acceleration sensor, a data cable, and a host computer. The three-axis acceleration sensor is placed on the elevator car, and the motion parameters of the elevator are collected when the elevator is running, For example, the elevator movement direction, speed, and acceleration, the parameters collected by the accelerometer are used as input, and the travel distance of the elevator in each journey (between two stops) is estimated through the extended Kalman filter algorithm. And return the estimated displacement and variance of the elevator during one trip. Combined with the SLAM algorithm, the position of the elevator car and the height of the floor are estimated (on the basis of setting the reference point). Finally, the position estimation is used to determine whether the elevator has abnormal parking failure.

本发明提出的一种基于三轴加速度传感器的电梯运动状态估计和停层异常检测方法，其步骤如下：A method for estimating the motion state of an elevator and detecting an abnormality of a floor stop based on a three-axis acceleration sensor proposed by the present invention, the steps of which are as follows:

1.电梯轿厢行驶距离估计1. Elevator car travel distance estimation

11)在电梯行驶距离估计中，电梯轿厢在连续两次停车(l-1,l)间的移动距离δp(l)和运动速度，通过横向和纵向速度为零的约束条件下辅助惯性导航即可估计得到。首先假设在t时刻，电梯的状态向量为xd(t)，其定义为11) In the estimation of the travel distance of the elevator, the moving distance δp(l) and the moving speed of the elevator car between two consecutive stops (l-1, l) are used to assist inertial navigation under the constraint that the lateral and longitudinal speeds are zero. can be estimated. First assume that at time t, the state vector of the elevator is x d (t), which is defined as

Figure GDA0002600768420000071
Figure GDA0002600768420000071

其中，d(t),s(t),δuz(t)分别表示电梯最后一次停车后运动的距离、轿厢的速度和垂直加速度测量偏差(包括地球引力造成的偏移)，s(t-1)为辅助状态变量。Among them, d(t), s(t), δu z (t) represent the distance traveled by the elevator after the last stop, the speed of the car and the measurement deviation of vertical acceleration (including the offset caused by the earth's gravity), s(t -1) is an auxiliary state variable.

12)在一次运动中，根据零速辅助惯性导航系统，可以定义时刻t电梯轿厢的运动状态矢量为：12) In a motion, according to the zero-speed auxiliary inertial navigation system, the motion state vector of the elevator car at time t can be defined as:

Figure GDA0002600768420000072
Figure GDA0002600768420000072

其中in

Figure GDA0002600768420000073
Figure GDA0002600768420000073

xd(t-1)是从上次停止开始行驶的距离，

Figure GDA0002600768420000074
是电梯轿厢测量的加速度，Δt为采样间隔，w(t)过程噪声(高斯白噪声，其协方差
Figure GDA0002600768420000075
其中
Figure GDA0002600768420000076
表示加速度计噪声方差，
Figure GDA0002600768420000077
表示加速度计偏差模型的噪声方差。x d (t-1) is the distance traveled since the last stop,
Figure GDA0002600768420000074
is the acceleration measured by the elevator car, Δt is the sampling interval, w(t) process noise (Gaussian white noise, its covariance
Figure GDA0002600768420000075
in
Figure GDA0002600768420000076
represents the accelerometer noise variance,
Figure GDA0002600768420000077
Represents the noise variance of the accelerometer bias model.

13)由于电梯轿厢的位置和运动状态只用加速度计来测量估计，因此无法为状态空间模型制定传统的量测更新方程。但是，电梯轿厢一般都是静止或者匀速运动，这两种状态都可以通过加速度计检测出来。假设已知电梯轿厢静止或者匀速运动的时间，那么就可以建立状态空间的伪量测模型：13) Since the position and motion state of the elevator car are only measured and estimated by the accelerometer, the traditional measurement update equation cannot be formulated for the state space model. However, the elevator car is generally stationary or moving at a constant speed, both of which can be detected by an accelerometer. Assuming that the time when the elevator car is stationary or moving at a constant speed is known, a pseudo-measurement model of the state space can be established:

Figure GDA0002600768420000078
Figure GDA0002600768420000078

其中

Figure GDA0002600768420000081
Figure GDA0002600768420000082
和
Figure GDA0002600768420000083
分别是匀速运动和静止时的观测噪声，即伪量测模型为电梯运动估计提供了两个不同的更新模型：零速更新模型和匀速更新模型。in
Figure GDA0002600768420000081
Figure GDA0002600768420000082
and
Figure GDA0002600768420000083
They are the observation noise when moving at a constant speed and at rest respectively, that is, the pseudo-measurement model provides two different update models for elevator motion estimation: a zero-speed update model and a constant-speed update model.

14)众所周知，电梯轿厢只是在有限范围内垂直方向移动。为了简化模型，方便计算，本发明将电梯轿厢的运动均视为匀速直线运动(静止是一种特殊的匀速直线运动)，同时三轴加速度传感器测量的加速度也取行程时间内的均值，并将电梯轿厢开始加速或减速的问题转化为电梯轿厢加速度均值局部变化的问题。14) It is well known that the elevator car only moves vertically within a limited range. In order to simplify the model and facilitate the calculation, the present invention regards the movement of the elevator car as uniform linear motion (stationary is a special uniform linear motion), and the acceleration measured by the three-axis acceleration sensor also takes the average value within the travel time, and The problem of starting to accelerate or decelerate the elevator car is transformed into a problem of local changes in the mean value of the elevator car acceleration.

15)根据状态空间矢量建立基于扩展卡尔曼滤波器的轿厢行驶距离估计和轿厢运动状态估计，将三轴加速度计测得平均加速度(或为负)

Figure GDA0002600768420000084
作为算法的输入。首先，通过扩展卡尔曼滤波算法对状态矢量估计值
Figure GDA0002600768420000085
和它的协方差矩阵
Figure GDA0002600768420000086
进行更新迭代，具体迭代步骤分为预测与更新：15) Establish car travel distance estimation and car motion state estimation based on extended Kalman filter according to the state space vector, and measure the average acceleration (or negative) with the three-axis accelerometer
Figure GDA0002600768420000084
as input to the algorithm. First, the state vector is estimated by the extended Kalman filter algorithm.
Figure GDA0002600768420000085
and its covariance matrix
Figure GDA0002600768420000086
The update iteration is performed, and the specific iteration steps are divided into prediction and update:

①预测①Prediction

Figure GDA0002600768420000087
Figure GDA0002600768420000087

Figure GDA0002600768420000088
Figure GDA0002600768420000088

②更新②Update

Figure GDA0002600768420000089
Figure GDA0002600768420000089

Figure GDA00026007684200000810
Figure GDA00026007684200000810

Figure GDA00026007684200000811
Figure GDA00026007684200000811

然后，通过均匀线性运动检测器判断电梯轿厢是否是均匀线性运动，如果是就继续执行迭代更新算法，如果不是就进行下一次行程的计算。Then, the uniform linear motion detector is used to judge whether the elevator car is in uniform linear motion, if so, continue to execute the iterative update algorithm, and if not, perform the calculation of the next trip.

16)电梯处于均匀线性运动时，通过计算将零速度假设拟合到观测数据后获得归一化预测误差ξd的大小，判断电梯轿厢是静止还是匀速运动。设置阈值γd，以便比较。确定电梯轿厢的状态，对观测模型进行更新：如果ξd＞γd，选择匀速测量更新；如果ξd＜γd，选择零速度测量更新。16) When the elevator is in uniform linear motion, the normalized prediction error ξ d is obtained by fitting the zero speed assumption to the observation data by calculation, and it is judged whether the elevator car is stationary or moving at a uniform speed. Set a threshold γd for comparison. Determine the state of the elevator car and update the observation model: if ξ d > γ d , choose constant speed measurement update; if ξ d < γ d , choose zero speed measurement update.

17)当算法检测到电梯轿厢停止运动并且行驶的距离超过了阈值γδp，就会输出电梯轿厢行驶的距离估计值

Figure GDA0002600768420000091
和其方差估计值
Figure GDA0002600768420000092
并且对系统状态矢量进行更新，并返回更新值，在下一次运动时重复15)-17)过程。17) When the algorithm detects that the elevator car stops moving and the distance traveled exceeds the threshold γδp , it outputs an estimate of the distance traveled by the elevator car
Figure GDA0002600768420000091
and its variance estimate
Figure GDA0002600768420000092
And update the system state vector, and return the updated value, repeat the process 15)-17) in the next movement.

2.电梯轿厢停层及故障判断2. Elevator car floor stop and fault judgment

21)通过第1阶段可以获取一系列的轿厢行驶距离估计值

Figure GDA0002600768420000093
和方差估计值
Figure GDA0002600768420000094
在此基础上，本发明通过SLAM算法可以实现电梯轿厢位置和停靠楼层估计，同时通过估计的轿厢位置，检测电梯是否不平层、蹲底或冲顶等故障。21) A series of car travel distance estimates can be obtained through stage 1
Figure GDA0002600768420000093
and variance estimates
Figure GDA0002600768420000094
On this basis, the present invention can realize the estimation of elevator car position and landing floor through the SLAM algorithm, and at the same time, through the estimated car position, it can detect whether the elevator is not leveling, squatting at the bottom or hitting the top and other faults.

22)引入新的状态矢量

Figure GDA0002600768420000095
其中，p(l)为电梯轿厢在第l趟行驶后的位置，m(i)为第i层楼的高度，M为楼层总数。由于每层楼的高度是恒定的，状态矢量可以描述为：22) Introduce a new state vector
Figure GDA0002600768420000095
Among them, p(l) is the position of the elevator car after the lth run, m (i) is the height of the i-th floor, and M is the total number of floors. Since the height of each floor is constant, the state vector can be described as:

Figure GDA0002600768420000096
Figure GDA0002600768420000096

其中e1为单位阵，

Figure GDA0002600768420000097
是电梯轿厢行驶距离估计值的误差，该误差假定为零均值、不相关且其方差为
Figure GDA0002600768420000098
where e1 is the identity matrix,
Figure GDA0002600768420000097
is the error in the estimate of the distance traveled by the elevator car, which is assumed to be zero-mean, uncorrelated, and whose variance is
Figure GDA0002600768420000098

23)假设电梯轿厢第l次行驶所停的楼层i已知的，则对应的伪观测模型为：23) Assuming that the floor i where the elevator car stops at the lth time is known, the corresponding pseudo-observation model is:

Figure GDA0002600768420000099
Figure GDA0002600768420000099

其中

Figure GDA00026007684200000910
vs(l)是伪测量误差，该测量误差是电梯轿厢停靠点位置和楼层高度之间微小偏差，主要是由于电梯控制系统不完善而导致的。该误差假定为零均值、不相关且其方差为
Figure GDA00026007684200000911
假如楼层和电梯轿厢停靠点是已知，则可以通过基于扩展卡尔曼滤波的SLAM跟踪电梯轿厢位置、估计楼层高度。in
Figure GDA00026007684200000910
v s (l) is the pseudo-measurement error, which is the slight deviation between the elevator car stop position and the floor height, mainly caused by the imperfect elevator control system. This error is assumed to have zero mean, uncorrelated and its variance is
Figure GDA00026007684200000911
Provided the floors and elevator car stops are known, the elevator car position can be tracked and the floor height estimated by SLAM based on extended Kalman filtering.

24)在实际工作环境中，楼层总数M和电梯轿厢的行驶顺序是未知的。假如电梯轿厢初始状态估计

Figure GDA00026007684200000912
和相应的协方差矩阵
Figure GDA00026007684200000913
是有效的，则可以通过基于最大似然数的数据关联实现电梯轿厢位置与估计楼层高度间的关联，该数据关联选择使归一化预测误差最小的观测模型，故通过最小化函数即可获得电梯轿厢所处的楼层高度：24) In the actual working environment, the total number of floors M and the running sequence of the elevator car are unknown. If the initial state of the elevator car is estimated
Figure GDA00026007684200000912
and the corresponding covariance matrix
Figure GDA00026007684200000913
is effective, then the association between the elevator car position and the estimated floor height can be realized through the data association based on the maximum likelihood. The data association selects the observation model that minimizes the normalized prediction error, so the function can be minimized Get the floor height of the elevator car:

Figure GDA0002600768420000101
Figure GDA0002600768420000101

25)要完成24中所说的步骤，必须要有理想的初始值。获取理想初始状态的方法，可以通过让电梯轿厢从底层依次运动到最高层，每层至少停留一次。25) To complete the steps mentioned in 24, it is necessary to have ideal initial values. The method to obtain the ideal initial state is to move the elevator car sequentially from the bottom floor to the highest floor, stopping at least once on each floor.

26)基于扩展卡尔曼滤波器的SLAM算法，将第1阶段获得的电梯轿厢行驶的距离估计值

Figure GDA0002600768420000102
和距离估计误差的方差估计值
Figure GDA0002600768420000103
作为输入，对状态向量
Figure GDA0002600768420000104
和状态协方差
Figure GDA0002600768420000105
进行更新迭代，具体迭代步骤分为预测与更新：26) Based on the SLAM algorithm of the extended Kalman filter, the estimated value of the distance traveled by the elevator car obtained in the first stage
Figure GDA0002600768420000102
and the variance estimate of the distance estimation error
Figure GDA0002600768420000103
As input, for the state vector
Figure GDA0002600768420000104
and state covariance
Figure GDA0002600768420000105
The update iteration is performed, and the specific iteration steps are divided into prediction and update:

①预测①Prediction

Figure GDA0002600768420000106
Figure GDA0002600768420000106

Figure GDA0002600768420000107
Figure GDA0002600768420000107

②更新②Update

Figure GDA0002600768420000108
Figure GDA0002600768420000108

Figure GDA0002600768420000109
Figure GDA0002600768420000109

Figure GDA00026007684200001010
Figure GDA00026007684200001010

27)计算归一化预测误差

Figure GDA00026007684200001011
并通过最小化函数求出电梯轿厢所处的楼层i。27) Calculate the normalized prediction error
Figure GDA00026007684200001011
And the floor i where the elevator car is located is obtained through the minimization function.

28)比较归一化误差ξs与假定阈值γs的大小；如果ξs＞γs，则表明估计误差过大，故电梯轿厢运行异常，此时状态向量不进行量测更新，并且报警提示电梯轿厢异常停靠故障；如果误差ξs＜γs，则证明电梯轿厢运行正常，对状态向量进行量测更新，并返回更新的值，进入下一轮行驶过程并重复步骤26)-28)。28) Compare the size of the normalized error ξ s and the assumed threshold γ s ; if ξ s >γ s , it indicates that the estimation error is too large, so the elevator car runs abnormally, at this time, the state vector is not measured and updated, and an alarm is issued Prompt the elevator car to stop abnormally; if the error ξ s < γ s , it proves that the elevator car is running normally, the state vector is measured and updated, and the updated value is returned, and the next round of driving process is entered and step 26)- 28).

1)下面结合图1和图2进行说明。以一座七层楼电梯为例，该电梯一侧安装有10个MPU-9150三轴加速度传感器的检测模块，用以获取电梯的三个方向加速度。将采集到的电梯垂直方向上的加速度作为输入，构建电梯的系统状态矢量xd。并根据扩展卡尔曼滤波进行状态矢量矩阵及协方差矩阵进行预测更新。1) The following description will be made with reference to FIG. 1 and FIG. 2 . Taking a seven-story elevator as an example, 10 MPU-9150 three-axis acceleration sensor detection modules are installed on one side of the elevator to obtain the three-direction acceleration of the elevator. Using the collected acceleration in the vertical direction of the elevator as input, construct the system state vector x d of the elevator. And the state vector matrix and covariance matrix are predicted and updated according to the extended Kalman filter.

2)扩展卡尔曼滤波更新后，判断电梯轿厢是否处于平稳线性运动。本发明简化电梯的加速过程，将电梯轿厢在加速或者减速过程中的加速度大小通过取均值的方式确定，并将电梯轿厢何时加速或减速的检测表述为检测加速度计测量的平均值中的局部变化的问题。平稳线性运动判断器判断如果电梯处于匀速线性运动，则判断这是一次有效的运动，反之就结束进程。2) After the extended Kalman filter is updated, it is judged whether the elevator car is in a smooth linear motion. The invention simplifies the acceleration process of the elevator, determines the acceleration of the elevator car during the acceleration or deceleration process by taking the average value, and expresses the detection of when the elevator car accelerates or decelerates as the average value measured by the detection accelerometer. the problem of local variation. The smooth linear motion judger judges that if the elevator is in a uniform linear motion, it is judged that it is a valid motion, otherwise, the process ends.

3)将平稳线性运动分为静止和匀速运动，通过计算将零速度假设拟合到观测数据时获得的归一化预测误差的大小

Figure GDA0002600768420000111
如果超过设定值，就判断电梯轿厢速度不为零，处于匀速运动，并采用匀速观测模型
Figure GDA0002600768420000112
否则就判断电梯速度为零，处于静止状态，并采用零速观测模型
Figure GDA0002600768420000113
3) Divide the stationary linear motion into stationary and uniform motion, and calculate the magnitude of the normalized prediction error obtained when fitting the zero-velocity assumption to the observed data
Figure GDA0002600768420000111
If it exceeds the set value, it is judged that the speed of the elevator car is not zero, it is moving at a constant speed, and a constant speed observation model is adopted.
Figure GDA0002600768420000112
Otherwise, it is judged that the elevator speed is zero, it is in a static state, and the zero-speed observation model is adopted.
Figure GDA0002600768420000113

4)在新的观测模型下对电梯状态矢量矩阵、增益矩阵和协方差矩阵进行扩展卡尔曼滤波更新。4) Under the new observation model, the extended Kalman filter is updated to the elevator state vector matrix, gain matrix and covariance matrix.

5)当电梯轿厢停止运动时，检查行驶距离

Figure GDA0002600768420000114
是否大于阈值γs，如果都满足条件，则输出电梯轿厢的行驶距离估计值
Figure GDA0002600768420000115
和它的估计方差
Figure GDA0002600768420000116
并对电梯系统状态矢量矩阵和其协方差矩阵进行更新，并返回其更新值。5) When the elevator car stops moving, check the travel distance
Figure GDA0002600768420000114
Whether it is greater than the threshold γ s , if all meet the conditions, output the estimated value of the travel distance of the elevator car
Figure GDA0002600768420000115
and its estimated variance
Figure GDA0002600768420000116
And update the elevator system state vector matrix and its covariance matrix, and return its updated value.

6)获得电梯行程参数

Figure GDA0002600768420000117
和
Figure GDA0002600768420000118
后，结合SLAM算法完成电梯正常停层位置的学习。首先定义新的电梯状态矢量xs(l)。将第1阶段获得的电梯轿厢行驶的距离估计值
Figure GDA0002600768420000119
和距离估计误差的方差估计值
Figure GDA00026007684200001110
作为输入，根据基于扩展卡尔曼滤波器的SLAM算法对状态向量
Figure GDA00026007684200001111
和状态协方差
Figure GDA00026007684200001112
进行更新迭代，获得
Figure GDA00026007684200001113
和其协方差矩阵Ps。6) Obtain elevator travel parameters
Figure GDA0002600768420000117
and
Figure GDA0002600768420000118
Then, combined with the SLAM algorithm, the learning of the normal stop position of the elevator is completed. First define a new elevator state vector x s (l). The estimated distance traveled by the elevator car obtained in stage 1
Figure GDA0002600768420000119
and the variance estimate of the distance estimation error
Figure GDA00026007684200001110
As input, according to the SLAM algorithm based on the extended Kalman filter, the state vector
Figure GDA00026007684200001111
and state covariance
Figure GDA00026007684200001112
Update iteratively to get
Figure GDA00026007684200001113
and its covariance matrix P s .

7)本发明采用基于最大似然数的数据关联，可以获得使归一化预测误差最小化的观测模型:

Figure GDA00026007684200001114
并通过最小化函数计算得到估计的电梯楼层高度和所在楼层i。7) The present invention adopts the data association based on maximum likelihood, and can obtain the observation model that minimizes the normalized prediction error:
Figure GDA00026007684200001114
And the estimated elevator floor height and floor i are obtained through the minimization function calculation.

8)如果预测误差的大小低于异常值阈值，则对新的状态矢量估计值

Figure GDA0002600768420000121
增益矩阵和协方差矩阵进行更新，否则调用异常停止程序，显示电梯异常位置停车。采用SLAM算法，电梯只需要完成一次正常运行(跑完所有的楼层)，系统就可以自主学习到全部正确的楼层位置。整个学习过程实现了与安装电梯大厦的楼层高度、楼层数，以及电梯品牌的无关性。8) If the magnitude of the prediction error is below the outlier threshold, estimate the value of the new state vector
Figure GDA0002600768420000121
The gain matrix and covariance matrix are updated, otherwise the abnormal stop procedure is called, and the elevator stops at the abnormal position. Using the SLAM algorithm, the elevator only needs to complete one normal operation (running all floors), and the system can autonomously learn all the correct floor positions. The whole learning process is independent of the floor height, the number of floors, and the elevator brand of the building where the elevator is installed.

9)如图3所示，电梯轿厢在安装三轴加速度传感器之后，所有传感器获取数据后通过扩展卡尔曼滤波估计得到的电梯位置估计值均为3σ置信区间内，说明滤波器估计合理且是一致估计。此外，经过开始的训练序列之后，基于SLAM的估计有效限制了位置误差增长，从滤波估计中很容易得到电梯行驶的具体楼层，而且当基于SLAM的估计检测到电梯轿厢停留在了两个楼层直接位置时，就会发出电梯轿厢异常停层的信号，说明本发明提出的方法不能能够构建电梯行驶的轨迹，还可以通过数据关联实现对电梯所处楼层的估计，并对电梯是否异常停层给出判断。9) As shown in Figure 3, after the three-axis acceleration sensor is installed in the elevator car, the estimated value of the elevator position obtained by the extended Kalman filter estimation after all the sensors obtain data are all within the 3σ confidence interval, indicating that the filter estimation is reasonable and is consensus estimate. In addition, after the initial training sequence, the SLAM-based estimation effectively limits the growth of the position error, and it is easy to obtain the specific floors where the elevator travels from the filtered estimation, and when the SLAM-based estimation detects that the elevator car stays on two floors When it is in the direct position, it will send a signal that the elevator car stops abnormally, indicating that the method proposed by the present invention cannot construct the trajectory of the elevator, and can also realize the estimation of the floor where the elevator is located through data association, and determine whether the elevator stops abnormally. layer to judge.

上述实施例为本发明较佳的实施方式，但本发明的实施方式并不受上述实施例的限制，其他的任何未背离本发明的精神实质与原理下所作的改变、修饰、替代、组合、简化，均应为等效的置换方式，都包含在本发明的保护范围之内。The above-mentioned embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the above-mentioned embodiments, and any other changes, modifications, substitutions, combinations, The simplification should be equivalent replacement manners, which are all included in the protection scope of the present invention.

Claims (5)

1.一种基于加速度传感器的电梯运动检测及异常位置停车判断方法，其特征在于按照如下步骤进行:1. a kind of elevator motion detection and abnormal position parking judgment method based on acceleration sensor, it is characterized in that carrying out according to the following steps: 步骤1.电梯轿厢行驶距离估计Step 1. Elevator car travel distance estimation 位于电梯轿厢上的三轴加速度传感器采集电梯的运动参数，通过扩展卡尔曼滤波算法，估计每一趟旅程中即两次停车之间电梯的行驶距离，返回电梯在一趟行驶过程中的位移估计值及方差；The three-axis acceleration sensor located on the elevator car collects the motion parameters of the elevator, and through the extended Kalman filter algorithm, the travel distance of the elevator between two stops in each trip is estimated, and the displacement of the elevator during one trip is returned. Estimates and variances; 步骤2.电梯轿厢停层及故障判断Step 2. Elevator car floor stop and fault judgment 基于步骤1的位移估计值及方差结果，结合SLAM算法，在设置基准点的基础上估计出电梯轿厢的位置和楼层的高度；Based on the displacement estimation value and variance result of step 1, combined with the SLAM algorithm, the position of the elevator car and the height of the floor are estimated on the basis of setting the reference point; 步骤3.通过位置估计判断电梯是否发生异常停车故障；所述的电梯轿厢行驶距离估计按照如下步骤进行：Step 3. Determine whether the elevator has an abnormal parking fault through position estimation; the estimation of the travel distance of the elevator car is carried out according to the following steps: 11)在电梯行驶距离估计中，电梯轿厢在连续两次停车(l-1,l)间移动距离为δp(l)和运动速度，通过横向和纵向速度为零的约束条件下辅助惯性导航即可估计得到，首先假设在t时刻，电梯的状态向量为xd(t)，其定义为11) In the estimation of the travel distance of the elevator, the moving distance of the elevator car between two consecutive stops (l-1,l) is δp(l) and the moving speed, and the inertial navigation is assisted under the constraint that the lateral and longitudinal speeds are zero. It can be estimated, first assume that at time t, the state vector of the elevator is x d (t), which is defined as
Figure FDA0002600768410000011
Figure FDA0002600768410000011
 其中，d(t),s(t),δuz(t)分别表示电梯最后一次停车后运动的距离、轿厢的速度和垂直加速度测量偏差，包括地球引力造成的偏移，s(t-1)为辅助状态变量；Among them, d(t), s(t), δu z (t) represent the distance traveled by the elevator after the last stop, the speed of the car and the measurement deviation of vertical acceleration, including the offset caused by the earth's gravity, s(t- 1) is an auxiliary state variable; 12)在一次运动中，根据零速辅助惯性导航系统，定义时刻t电梯轿厢的运动状态矢量为：12) In a motion, according to the zero-speed auxiliary inertial navigation system, the motion state vector of the elevator car at time t is defined as:
Figure FDA0002600768410000012
Figure FDA0002600768410000012
 其中in
Figure FDA0002600768410000013
Figure FDA0002600768410000013
 xd(t-1)是从上次停止开始行驶的距离，
Figure FDA0002600768410000014
是电梯轿厢测量的加速度，Δt为采样间隔，w(t)过程噪声，高斯白噪声，其协方差
Figure FDA0002600768410000015
其中
Figure FDA0002600768410000016
表示加速度计噪声方差，
Figure FDA0002600768410000021
表示加速度计偏差模型的噪声方差；x d (t-1) is the distance traveled since the last stop,
Figure FDA0002600768410000014
is the acceleration measured by the elevator car, Δt is the sampling interval, w(t) process noise, Gaussian white noise, and its covariance
Figure FDA0002600768410000015
in
Figure FDA0002600768410000016
represents the accelerometer noise variance,
Figure FDA0002600768410000021
represents the noise variance of the accelerometer bias model; 13)由于电梯轿厢的位置和运动状态只用加速度计来测量估计，因此无法为状态空间模型制定传统的量测更新方程；但是，电梯轿厢一般都是静止或者匀速运动，这两种状态都通过加速度计检测出来；假设已知电梯轿厢静止或者匀速运动的时间，那么状态空间的伪量测模型如下：13) Since the position and motion state of the elevator car are only measured and estimated by the accelerometer, it is impossible to formulate a traditional measurement update equation for the state space model; however, the elevator car is generally stationary or moving at a uniform speed, and these two states All are detected by the accelerometer; assuming that the time when the elevator car is stationary or moving at a constant speed is known, the pseudo-measurement model of the state space is as follows:
Figure FDA0002600768410000022
Figure FDA0002600768410000022
 其中
Figure FDA0002600768410000023
Figure FDA0002600768410000024
和
Figure FDA0002600768410000025
分别是匀速运动和静止时的观测噪声，即伪量测模型为电梯运动估计提供了两个不同的更新模型：零速更新模型和匀速更新模型；in
Figure FDA0002600768410000023
Figure FDA0002600768410000024
and
Figure FDA0002600768410000025
They are the observation noise when moving at a constant speed and at rest, that is, the pseudo-measurement model provides two different update models for elevator motion estimation: a zero-speed update model and a constant-speed update model; 14)电梯轿厢只在有限范围内垂直方向移动，简化模型将电梯轿厢的运动均视为匀速直线运动，静止是一种特殊的匀速直线运动，同时三轴加速度传感器测量的加速度也取行程时间内的均值，并将电梯轿厢开始加速或减速的问题转化为电梯轿厢加速度均值局部变化的问题；14) The elevator car only moves in the vertical direction within a limited range. The simplified model regards the movement of the elevator car as a uniform linear motion. Rest is a special uniform linear motion. At the same time, the acceleration measured by the three-axis acceleration sensor is also taken as the stroke. The average value in time, and the problem of the elevator car starting to accelerate or decelerate is transformed into the problem of the local change of the average value of the elevator car acceleration; 15)根据状态空间矢量建立基于扩展卡尔曼滤波器的轿厢行驶距离估计和轿厢运动状态估计，将三轴加速度计测得平均加速度
Figure FDA0002600768410000026
作为算法的输入；15) Establish car travel distance estimation and car motion state estimation based on extended Kalman filter according to the state space vector, and measure the average acceleration of the three-axis accelerometer
Figure FDA0002600768410000026
as input to the algorithm; 16)电梯处于均匀线性运动时，通过计算将零速度假设拟合到观测数据后获得归一化预测误差ξd的大小，判断电梯轿厢是静止还是匀速运动；设置阈值γd，确定电梯轿厢的状态，对观测模型进行更新：如果ξd＞γd，选择匀速测量更新；如果ξd＜γd，选择零速度测量更新；16) When the elevator is in uniform linear motion, the normalized prediction error ξ d is obtained by fitting the zero-speed assumption to the observation data, and the elevator car is determined to be stationary or moving at a constant speed; set the threshold γ d to determine the elevator car According to the state of the car, update the observation model: if ξ d > γ d , select uniform speed measurement update; if ξ d < γ d , select zero speed measurement update; 17)当算法检测到电梯轿厢停止运动并且行驶的距离超过了阈值γδp，就会输出电梯轿厢行驶的距离估计值
Figure FDA0002600768410000027
和其方差估计值
Figure FDA0002600768410000028
并且对系统状态矢量进行更新，并返回更新值，在下一次运动时重复15)-17)过程。17) When the algorithm detects that the elevator car stops moving and the distance traveled exceeds the threshold γδp , it outputs an estimate of the distance traveled by the elevator car
Figure FDA0002600768410000027
and its variance estimate
Figure FDA0002600768410000028
And update the system state vector, and return the updated value, repeat the process 15)-17) in the next movement. 2.根据权利要求1所述的基于加速度传感器的电梯运动检测及异常位置停车判断方法，其特征在于：步骤15)中2. the elevator motion detection and abnormal position parking judgment method based on acceleration sensor according to claim 1, is characterized in that: in step 15) 首先，通过扩展卡尔曼滤波算法对状态矢量估计值
Figure FDA0002600768410000029
和它的协方差矩阵
Figure FDA0002600768410000031
进行更新迭代，具体迭代步骤分为预测与更新：First, the state vector is estimated by the extended Kalman filter algorithm.
Figure FDA0002600768410000029
and its covariance matrix
Figure FDA0002600768410000031
The update iteration is performed, and the specific iteration steps are divided into prediction and update: ①预测①Prediction
Figure FDA0002600768410000032
Figure FDA0002600768410000032
Figure FDA0002600768410000033
Figure FDA0002600768410000033
 ②更新②Update
Figure FDA0002600768410000034
Figure FDA0002600768410000034
Figure FDA0002600768410000035
Figure FDA0002600768410000035
Figure FDA0002600768410000036
Figure FDA0002600768410000036
 然后，通过均匀线性运动检测器判断电梯轿厢是否是均匀线性运动，如果是就继续执行迭代更新算法，如果不是就进行下一次行程的计算。Then, the uniform linear motion detector is used to judge whether the elevator car is in uniform linear motion, if so, continue to execute the iterative update algorithm, and if not, perform the calculation of the next trip. 3.根据权利要求1所述的基于加速度传感器的电梯运动检测及异常位置停车判断方法，其特征在于所述的步骤2电梯轿厢停层及故障判断按照如下步骤进行：3. the elevator motion detection and abnormal position parking judgment method based on acceleration sensor according to claim 1 is characterized in that described step 2 elevator car stops floor and fault judgment is carried out according to the following steps: 21)通过步骤1获取轿厢行驶距离估计值
Figure FDA0002600768410000037
和方差估计值
Figure FDA0002600768410000038
在此基础上，通过SLAM算法实现电梯轿厢位置和停靠楼层估计，同时通过估计的轿厢位置，检测电梯是否存在故障；21) Obtain the estimated value of the car travel distance through step 1
Figure FDA0002600768410000037
and variance estimates
Figure FDA0002600768410000038
On this basis, the SLAM algorithm is used to estimate the elevator car position and the parking floor, and at the same time, the estimated car position is used to detect whether the elevator is faulty; 22)引入新的状态矢量
Figure FDA0002600768410000039
其中，p(l)为电梯轿厢在第l趟行驶后的位置，δp(l)表示电梯轿厢在连续两次停车(l-1,l)间的移动距离；m(i)为第i层楼的高度，M为楼层总数；由于每层楼的高度是恒定的，状态矢量描述为：22) Introduce a new state vector
Figure FDA0002600768410000039
Among them, p(l) is the position of the elevator car after the lth trip, δp(l) represents the moving distance of the elevator car between two consecutive stops (l-1,l); m (i) is the first The height of the i floor, M is the total number of floors; since the height of each floor is constant, the state vector is described as:
Figure FDA00026007684100000312
Figure FDA00026007684100000312
 其中e1为单位阵，
Figure FDA00026007684100000310
是电梯轿厢行驶距离估计值的误差，假定误差ωs(l)为零均值、不相关，且其方差为
Figure FDA00026007684100000311
where e1 is the identity matrix,
Figure FDA00026007684100000310
is the error of the estimated value of the travel distance of the elevator car, assuming that the error ω s (l) is zero mean, uncorrelated, and its variance is
Figure FDA00026007684100000311
 23)假设电梯轿厢第l次行驶所停的楼层i已知的，则对应的伪观测模型为：23) Assuming that the floor i where the elevator car stops at the lth time is known, the corresponding pseudo-observation model is:
Figure FDA0002600768410000041
Figure FDA0002600768410000041
 其中
Figure FDA0002600768410000042
vs(l)是伪测量误差，由于电梯控制系统不完善该伪测量误差是电梯轿厢停靠点位置和楼层高度之间微小偏差，该伪测量误差假定为零均值、不相关，伪测量误差vs(l)的方差为
Figure FDA0002600768410000043
假如楼层和电梯轿厢停靠点是已知，则通过基于扩展卡尔曼滤波的SLAM跟踪电梯轿厢位置、估计楼层高度；in
Figure FDA0002600768410000042
v s (l) is a pseudo-measurement error. Due to the imperfect elevator control system, the pseudo-measurement error is a small deviation between the elevator car stop position and the floor height. The pseudo-measurement error is assumed to have zero mean and no correlation. The pseudo-measurement error The variance of v s (l) is
Figure FDA0002600768410000043
If the floor and elevator car stops are known, the elevator car position is tracked and the floor height is estimated by SLAM based on extended Kalman filtering; 24)在实际工作环境中，楼层总数M和电梯轿厢的行驶顺序是未知的；假如电梯轿厢初始状态估计
Figure FDA0002600768410000044
和相应的协方差矩阵
Figure FDA0002600768410000045
是有效的，则通过基于最大似然数的数据关联实现电梯轿厢位置与估计楼层高度间的关联，该数据关联选择使归一化预测误差最小的观测模型，故通过最小化函数即可获得电梯轿厢所处的楼层高度：24) In the actual working environment, the total number of floors M and the travel sequence of the elevator car are unknown; if the initial state of the elevator car is estimated
Figure FDA0002600768410000044
and the corresponding covariance matrix
Figure FDA0002600768410000045
is valid, then the association between the elevator car position and the estimated floor height is realized through the data association based on the maximum likelihood. The data association selects the observation model that minimizes the normalized prediction error, so it can be obtained by minimizing the function. Floor height of the elevator car:
Figure FDA0002600768410000046
Figure FDA0002600768410000046
 25)要完成上述24)中所说的步骤，必须要有理想的初始值；获取理想初始状态的方法，通过让电梯轿厢从底层依次运动到最高层，每层至少停留一次；25) To complete the steps mentioned in the above 24), there must be an ideal initial value; the method for obtaining the ideal initial state is to allow the elevator car to move from the bottom floor to the highest floor in sequence, and stay at least once on each floor; 26)基于扩展卡尔曼滤波器的SLAM算法，将第1阶段获得的电梯轿厢行驶的距离估计值
Figure FDA0002600768410000047
和距离估计误差的方差估计值
Figure FDA0002600768410000048
作为输入，对状态向量
Figure FDA00026007684100000411
和状态协方差
Figure FDA0002600768410000049
进行更新迭代；26) Based on the SLAM algorithm of the extended Kalman filter, the estimated value of the distance traveled by the elevator car obtained in the first stage
Figure FDA0002600768410000047
and the variance estimate of the distance estimation error
Figure FDA0002600768410000048
As input, for the state vector
Figure FDA00026007684100000411
and state covariance
Figure FDA0002600768410000049
update iteratively; 27)计算归一化预测误差
Figure FDA00026007684100000410
并通过最小化函数求出电梯轿厢所处的楼层i；27) Calculate the normalized prediction error
Figure FDA00026007684100000410
And find the floor i where the elevator car is located by minimizing the function; 28)比较归一化误差ξs与假定阈值γs的大小，如果ξs＞γs，则表明估计误差过大，故电梯轿厢运行异常，此时状态向量不进行量测更新，并且报警提示电梯轿厢异常停靠故障；如果误差ξs＜γs，则证明电梯轿厢运行正常，对状态向量进行量测更新，并返回更新的值，进入下一轮行驶过程并重复步骤26)-28)。28) Compare the size of the normalized error ξ s and the assumed threshold γ s , if ξ s >γ s , it means that the estimation error is too large, so the elevator car runs abnormally, at this time, the state vector is not measured and updated, and an alarm is issued Prompt the elevator car to stop abnormally; if the error ξ s < γ s , it proves that the elevator car is running normally, measure and update the state vector, and return the updated value, enter the next round of driving process and repeat step 26)- 28). 4.根据权利要求3所述的基于加速度传感器的电梯运动检测及异常位置停车判断方法，其特征在于所述的步骤26)，具体迭代步骤分为预测与更新：4. the elevator motion detection and abnormal position parking judgment method based on acceleration sensor according to claim 3, it is characterized in that described step 26), concrete iteration step is divided into prediction and update: ①预测①Prediction
Figure FDA0002600768410000051
Figure FDA0002600768410000051
Figure FDA0002600768410000052
Figure FDA0002600768410000052
 ②更新②Update
Figure FDA0002600768410000053
Figure FDA0002600768410000053
Figure FDA0002600768410000054
Figure FDA0002600768410000054
Figure FDA0002600768410000055
Figure FDA0002600768410000055
 5.根据权利要求3所述的基于加速度传感器的电梯运动检测及异常位置停车判断方法，其特征在于：检测电梯是否存在故障，故障包括不平层、蹲底或冲顶。5. The acceleration sensor-based elevator motion detection and abnormal position parking judgment method according to claim 3, characterized in that: detecting whether there is a fault in the elevator, and the fault includes uneven floor, squatting bottom or topping.
CN201811112016.6A 2018-09-25 2018-09-25 Elevator motion detection and abnormal position parking judgment method based on acceleration sensor Active CN109211233B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811112016.6A CN109211233B (en) 2018-09-25 2018-09-25 Elevator motion detection and abnormal position parking judgment method based on acceleration sensor

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811112016.6A CN109211233B (en) 2018-09-25 2018-09-25 Elevator motion detection and abnormal position parking judgment method based on acceleration sensor

Publications (2)

Publication Number Publication Date
CN109211233A CN109211233A (en) 2019-01-15
CN109211233B true CN109211233B (en) 2020-10-23

Family

ID=64984602

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811112016.6A Active CN109211233B (en) 2018-09-25 2018-09-25 Elevator motion detection and abnormal position parking judgment method based on acceleration sensor

Country Status (1)

Country Link
CN (1) CN109211233B (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109748170B (en) * 2019-01-18 2020-05-26 西人马帝言(北京)科技有限公司 Elevator fault warning method, device, equipment and medium
CN109978026B (en) * 2019-03-11 2021-03-09 浙江新再灵科技股份有限公司 Elevator position detection method and system based on LSTM network
CN110135241B (en) * 2019-03-27 2021-08-31 浙江新再灵科技股份有限公司 Statistical analysis system of elevator stroke
CN112047209B (en) * 2020-09-09 2022-09-13 上海有个机器人有限公司 Automatic calibration method, medium, terminal and device for elevator floors
CN113277392B (en) * 2021-04-28 2024-02-02 上海有个机器人有限公司 Elevator floor height calibration method based on Gaussian distribution model, electronic equipment and storage medium
CN114357794B (en) * 2022-01-12 2024-07-05 东南大学 Bus arrival and stop evaluation method based on improved Kalman filtering
CN114560360B (en) * 2022-02-10 2024-11-01 北京三快在线科技有限公司 Method and device for taking elevator by robot, medium and robot controller
CN116969286B (en) * 2023-09-20 2023-12-26 杭州海康威视数字技术股份有限公司 Floor positioning method and device
CN117985559B (en) * 2024-04-02 2024-06-28 深圳市博安智控科技有限公司 Elevator management method and equipment applied to building

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102633173B (en) * 2012-05-04 2014-10-15 林创鲁 System and method for monitoring operation state of elevator car
US9235657B1 (en) * 2013-03-13 2016-01-12 Johnson Controls Technology Company System identification and model development
US9374678B2 (en) * 2014-02-28 2016-06-21 Texas Instruments Incorporated Kalman filter iteratively performing time/measurement update of user/relative floor locations
CN104316060B (en) * 2014-06-06 2018-09-11 清华大学深圳研究生院 The spacecrafts rendezvous method and device of space non-cooperative target
CN104062977B (en) * 2014-06-17 2017-04-19 天津大学 Full-autonomous flight control method for quadrotor unmanned aerial vehicle based on vision SLAM
CN106441309B (en) * 2016-11-14 2019-03-12 郑州轻工业学院 Distributed autonomous navigation method for Mars entry segment based on covariance cross fusion
CN107215734A (en) * 2017-07-06 2017-09-29 天津康途科技有限公司 A kind of method and system detected for elevator real time acceleration and speed and position

Also Published As

Publication number Publication date
CN109211233A (en) 2019-01-15

Similar Documents

Publication Publication Date Title
CN109211233B (en) Elevator motion detection and abnormal position parking judgment method based on acceleration sensor
CN108140052B (en) Computing system with structural vertical hierarchy mechanism and method of operation thereof
CN107990895B (en) A method and system for pedestrian trajectory tracking between floors in buildings based on wearable IMU
CN108957510B (en) Pedestrian seamless combined navigation and positioning method based on inertia/zero speed/GPS
JP6462692B2 (en) Autonomous mobile device
US12202705B2 (en) Estimation and presentation of area of interest for condition based monitoring on doors: door health heat map
CN104781173B (en) Inertial measurement unit assisted elevator position calibration
EP4414749A1 (en) Data processing method and apparatus
CN109708630B (en) Step strapdown height measurement method based on SHE model
CN113295158A (en) Indoor positioning method fusing inertial data, map information and pedestrian motion state
CN108168509A (en) A kind of quadrotor Error Tolerance method of estimation of lift model auxiliary
CN110346821A (en) A kind of SINS/GPS integrated attitude determination localization method solving the problems, such as GPS long-time losing lock and system
CN110555398A (en) Fault diagnosis method for determining first arrival moment of fault based on optimal filtering smoothness
CN114265084A (en) Unmanned vehicle dynamic obstacle avoidance method based on particle filtering
CN109978026B (en) Elevator position detection method and system based on LSTM network
US20210087018A1 (en) Air pressure floor table detection: statistical analysis of location
CN113359809B (en) Bridge detection unmanned aerial vehicle autonomous positioning method based on RBFNN assistance
Rudin et al. A sensor fault detection for aircraft using a single Kalman filter and hidden Markov models
CN108387233B (en) A Pedestrian Motion Pattern Judgment Method Based on Fuzzy Logic
CN110968099B (en) Robot trapped detection method and robot
CN112947420B (en) Equipment running state identification method, device, robot and storage medium
CN106896361A (en) A kind of deep water robot multi-model EKF combined navigation devices and method
CN111693066B (en) Slip identification and intelligent compensation method for mine heading machine
CN116659496A (en) Robot navigation positioning system and method combining laser range finder and accelerometer
CN115248037A (en) Intelligent vehicle cooperative positioning method combining distance measurement, direction measurement and speed measurement

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant