Crushing Circuit Optimisation

End-to-End AI Solutions

Plant Data Analytics

All our solutions are based on plant data that helps us benchmark your plant's performance, identify your bottlenecks and rank the opportunities. Through our holistic approach we evaluate the overall alarm system performance, downtime logs and correlate those with availalble process data and the control system scheme.
Computer Vision (CV)

We utilise state of the art convolutional neural network based computer vision algorithms that are custom designed to your specific application. We offer detailed camera design, sensor fusion recommendation and edge hardware. Our CV software can detect product quality and granularity as well as dynamic events like blockages, liquefaction, contamination etc.
Control System Design

Through automation, Espy reduces the workload on mine operators so they can focus on higher level tasks. Alarm system performance optimisation, control loop performance improvement, high level control scheme optimisation are a few examples of the experties of the Espy team.

ROM Bin Blockage Detection and Truck Control

Our CV system can detect blockages of static grizzlies/crushers and control the traffic lights/trucks to prevent serious bogging. The system can initiate rock breaking or warn operators when of the blockages. The system has also the capability to recognise granularity and grade information that can be fed forward to the plant.
Apron Feeder Control

We offer AI model based Apron feeder control to tackle surging, large transfer delay and product characteristic changes. Our sensor fusion based systems offer unprecedented accuracy of detecting surges on the Apron feeder.
Crusher Choke Control and CSS Optimisation

Our CV solutions can be fed forward to dynamically adjust the Closed Side Setting of the crusher. With sensor fusion using AI, we can complement existing crusher bowl level control with out CV solution and offer a more robust choke control and better utilisation of crushers.

Espy combines the power of computer vision and sensor fusion based material characteristic detection with advanced process control schemes.

AI Model Predictive Feeder Control

Reinforcement Learning based Feeder Discharge Models incorporate equipment parameters, ROM bin level, ore characteristic and PSD to provide an accurate representation of the discharege flow. Espy models are trained on real, process data and engineered to the specific plant design, so they can operate at maximum efficiency.
Crusher Optimal Load Balancing

Computer vision based ore size and blockage detection helps prevent downtime and optimises crusher utilisation. Espy' high accuracy, semantic, AI models detect blockages real-time in harsh environment. The models can be deployed on premise or hosted from the Espy cloud. Maintenance and fine-tuning of the model can also be done remotely or on a locally hosted system.

Truck Control

The Espy Computer Vision System detects the location and extent of blockages on the ROM Bin grizzly. The blocakges weighed based on their location so the performance of the grizzly can be evaluated and fed back to the process control system (PCS). The PCS via the fleet management system decides if trucks can continue tipping or have to stop reversing to the ROM so unblocking can commence.
Improved Rock Breaking Repsonse Time

As the blockage on the static grizzly evolves the operator gets notified with increasing priority alarms so necessary measures can be taken. The Espy Computer Vision System interacts with the operator and it records those operator decisions that differ from its own so those can be used for further training and fine tuning.
Blend Control

If material is trucked from multiple sources, the system can integrate with the blend control to feed finer material while preparation for unblocking is undergoing.
Autonomous Rock Breaking

With the exact knowledge of the location of the blockages autonomous rock breaker positioning and rock breaking can take place. The Espy Computer Vision System provides an open API for the integration of any rock breakers.
Oversize Metics

The CV system has the added benefit that the change in blocked area gives a truck by truck indication of the oversize material, which can be fed back to the mine for geology model and blasting optimisation

Crusher Bridge Prediction

The Espy Computer Vision System is able to detect dynamic bridge formations using state of the art video analytics, where multiple frames are analysed with sequence models to give the fastest prediction of potential bridge formations.
Optimal Crusher Feed Control

With high accuracy blocakge detection, the Espy Computer Vision System allows maximum utilisation of crushers without risking catastrophic bogging. The system initiates tipping as soon as no blockage has been confirmed and also with the optimal level of material left in the crusher bowl for minimising wear.
Dynamic Closed Side Setting Control

The Espy Computer Vision System recognises the granularity of the individual truck tips and dynamically feds forward the information to the CSS control. The control can adapt to any variability in granularity truck by truck basis.
Conveyor Volume Limit Feed Forward

Combining the granularity with the dynamic CSS control the Espy system can feed forward the volume constraints to the conveying system as the material granularity changes allowing for maximum utilisation of the conveyors and increased throughput.
Grade Recognition

While material being crushed the Espy Computer Vision System can recognise material grade to feed it back to grade control, geology and stockyard control.

Crusher Bowl Level Detection with Sensor Fusion

The Espy Computer Vision System can complement conventional level sensors of any type (radar, ultrasonic or laser) and with a sensor fusion algorithm the system can automatically fall back to the conventional level sensors when visibility is compromised.
Real-time Crusher Bowl Volume Control

State-of-the-art convolutional neural networks evaluate the video stream at high speed (sub second) so the bowl volume can be fed back to the apron feeder control scheme and used for real-time constraint control.
Blockage Detection

Multi-frame video analytics is used to detect blockages even in the ore stream so these can be used to cut feed and prevent from more serious bogging.
Ore Feed Moisture Content Recognition

The CV system is capable of recognising changes in moisture content that can be fed forward to the ore conditioning and dust suppression systems, which would improve health and environmental conditions as well as product quality.